QA report: support-intents

The data-quality report on this adapter's training, development, calibration and test files, written by the maintainers' QA script before training and reviewed by someone who did not build the data. For publication, internal file paths were cut to file names and machine, service and account names were removed; every number, verdict and sample row is unchanged.

QA: support-intents

Checked 2026-09-30 12:05 by adapters/qa/qa.py (READY file ready-support-intents-v2.json, v2 (QA fixes 2026-09-30)).

Verdict: PASS WITH NOTES

support-intents v2 (READY line 2026-09-30 11:59; train sha256 82047bf5…7ca2). The fixes for the v1 findings (in notes-support-intents-v1.md) are confirmed.

v1 findings, now fixed (checked)

Unfilled Bitext slots: none are left in any split (v1 had 3,586 in train). Each slot is filled with a typed random value, for example "invoice #85632", "purchase ORD-784868" or "standard account".
Near duplicates across splits: clusters were formed before splitting. 0% of dev and calibration rows, and 0.1% of test rows, have a near duplicate in train (v1: 3–5%). Within train, 2.5% of rows are excess near duplicates (Bitext paraphrases).

Notes (for the model card and v3; no shortcut)

Filled value formats follow the slot type. "#digits" appears only in invoice intents (894 rows, split about evenly between check_invoice and get_invoice), and "ORD-…" only in order intents (440 rows across change, track and cancel). Each format is shared within its group of intents, so neither points to one label, and the words "invoice" and "order" are always present anyway. A message like "track order #123" is still a small risk. v3: draw ID formats from one pool for all slot types.
Style follows the source data set. HWU64 and SNIPS texts are lowercase with no end punctuation. Each data set has its own options and its own service text, so this cannot pick a wrong option.
A surface-only model reaches 13.3% balanced accuracy on about 100 classes (chance 1%). That comes from the option count, prompt length and the data set, which the service field already names, so it is the task's structure. The option picker is at chance.
Sample reading: 25 random Bitext rows were natural (typos, swearing and casual phrasing kept from the source), and every label matched the text.

Automatic flags (for the reviewer to judge; not all are problems)

formatting 'ends with ?' differs by class: delivery_options 32%, contact_customer_service 30%, get_weather 27%, check_cancellation_fee 27% …
formatting 'ends with .' differs by class: general_repeat 100%, general_praise 100%, general_negate 100%, general_explain 100% …
formatting 'no end punctuation' differs by class: weather_query 100%, transport_traffic 100%, transport_ticket 100%, transport_taxi 100% …
formatting 'starts lowercase' differs by class: transport_traffic 100%, transport_taxi 100%, takeaway_query 100%, takeaway_order 100% …
formatting 'all lowercase' differs by class: transport_taxi 100%, takeaway_query 100%, takeaway_order 100%, qa_stock 100% …
formatting 'has a digit' differs by class: change_order 99%, track_order 98%, cancel_order 97%, rate_book 78% …
2552 strong phrase flags (see list): review whether they are meaning or leakage
60 standard phrase flags (≥2% of a class, mostly that class): review

Data checked

split	rows	families	file
train	54,849	54775	`train.jsonl`
dev	1,000	1000	`dev.jsonl`
calibration	999	999	`calibration.jsonl`
test	5,577	5576	`test.jsonl`

Train sha256: 82047bf50cb306b232624278de34ee1164f654a1f1e420a636b2f632ff317ca2 (READY file gives no checksum)
Main text field (the text the phrase and length checks use): state.message.
Label classes: add_to_playlist, alarm_query, alarm_remove, alarm_set, audio_volume_down, audio_volume_mute, audio_volume_up, book_restaurant, calendar_query, calendar_remove, calendar_set, cancel_order, change_order, change_shipping_address, check_cancellation_fee, check_invoice, check_payment_methods, check_refund_policy, complaint, contact_customer_service, contact_human_agent, cooking_recipe, create_account, datetime_convert, datetime_query, delete_account, delivery_options, delivery_period, edit_account, email_addcontact, email_query, email_querycontact, email_sendemail, general_affirm, general_commandstop, general_confirm, general_dontcare, general_explain, general_joke, general_negate, general_praise, general_quirky, general_repeat, get_invoice, get_refund, get_weather, iot_cleaning, iot_coffee, iot_hue_lightchange, iot_hue_lightdim, iot_hue_lightoff, iot_hue_lighton, iot_hue_lightup, iot_wemo_off, iot_wemo_on, lists_createoradd, lists_query, lists_remove, music_likeness, music_query, music_settings, news_query, newsletter_subscription, payment_issue, place_order, play_audiobook, play_game, play_music, play_podcasts, play_radio, qa_currency, qa_definition, qa_factoid, qa_maths, qa_stock, rate_book, recommendation_events, recommendation_locations, recommendation_movies, recover_password, registration_problems, review, search_creative_work, search_screening_event, set_up_shipping_address, social_post, social_query, switch_account, takeaway_order, takeaway_query, track_order, track_refund, transport_query, transport_taxi, transport_ticket, transport_traffic, weather_query (<listed option> = one of the per-row listed options such as t3 or o12). Row kinds (source.kind): intent.

3. Balance

Label class share per split

lclass	train	dev	calibration	test	train rows
add_to_playlist	3.4%	2.8%	4.3%	1.8%	1,863
alarm_query	0.3%	0.4%	0.5%	0.4%	160
alarm_remove	0.2%	0.0%	0.0%	0.2%	101
alarm_set	0.4%	0.2%	0.4%	0.5%	242
audio_volume_down	0.1%	0.1%	0.1%	0.2%	66
audio_volume_mute	0.2%	0.0%	0.1%	0.3%	131
audio_volume_up	0.2%	0.7%	0.2%	0.1%	121
book_restaurant	3.5%	2.8%	3.5%	1.8%	1,904
calendar_query	1.5%	1.5%	1.8%	1.5%	842
calendar_remove	0.8%	0.5%	0.5%	1.0%	435
calendar_set	2.2%	2.0%	2.3%	2.7%	1,186
cancel_order	0.7%	0.2%	0.5%	0.5%	388
change_order	1.4%	1.7%	1.1%	1.3%	744
change_shipping_address	1.5%	1.4%	1.5%	1.7%	833
check_cancellation_fee	1.5%	1.2%	1.2%	1.6%	815
check_invoice	1.4%	1.5%	1.4%	1.6%	791
check_payment_methods	1.6%	1.3%	1.7%	1.9%	854
check_refund_policy	1.5%	0.9%	0.8%	2.2%	829
complaint	1.6%	1.6%	0.8%	1.9%	867
contact_customer_service	1.6%	0.6%	1.4%	1.5%	870
contact_human_agent	1.5%	2.5%	1.8%	1.7%	833
cooking_recipe	0.6%	0.7%	1.0%	0.8%	315
create_account	1.4%	1.2%	2.1%	1.5%	743
datetime_convert	0.1%	0.3%	0.3%	0.1%	74
datetime_query	0.9%	1.3%	0.7%	1.1%	504
delete_account	1.4%	1.5%	1.3%	1.6%	776
delivery_options	1.0%	1.2%	0.9%	1.1%	558
delivery_period	1.6%	2.1%	1.0%	1.8%	853
edit_account	1.2%	1.1%	1.5%	1.4%	653
email_addcontact	0.1%	0.0%	0.0%	0.2%	74
email_query	1.1%	1.0%	0.7%	1.2%	625
email_querycontact	0.3%	0.4%	0.4%	0.4%	171
email_sendemail	1.0%	1.3%	0.9%	1.3%	554
general_affirm	0.9%	0.5%	0.8%	1.2%	469
general_commandstop	0.5%	0.5%	0.0%	0.7%	275
general_confirm	0.9%	0.8%	0.9%	1.0%	480
general_dontcare	0.7%	0.4%	0.9%	1.0%	379
general_explain	1.1%	1.0%	1.2%	1.1%	598
general_joke	0.2%	0.3%	0.1%	0.2%	101
general_negate	1.5%	1.3%	1.5%	1.9%	807
general_praise	1.2%	1.1%	1.2%	1.4%	683
general_quirky	1.6%	2.7%	2.2%	2.1%	866
general_repeat	0.9%	1.5%	0.7%	1.1%	500
get_invoice	1.4%	1.9%	1.1%	1.7%	753
get_refund	1.4%	2.3%	1.7%	1.5%	781
get_weather	3.5%	2.9%	3.7%	1.8%	1,919
iot_cleaning	0.2%	0.6%	0.2%	0.4%	129
iot_coffee	0.3%	0.3%	0.4%	0.4%	158
iot_hue_lightchange	0.3%	0.3%	0.0%	0.3%	186
iot_hue_lightdim	0.2%	0.2%	0.0%	0.2%	108
iot_hue_lightoff	0.3%	0.9%	0.3%	0.5%	191
iot_hue_lighton	0.1%	0.0%	0.1%	0.1%	30
iot_hue_lightup	0.2%	0.1%	0.0%	0.3%	113
iot_wemo_off	0.1%	0.1%	0.1%	0.1%	82
iot_wemo_on	0.1%	0.1%	0.0%	0.1%	68
lists_createoradd	0.4%	0.2%	0.7%	0.5%	237
lists_query	0.5%	0.9%	0.6%	0.5%	297
lists_remove	0.4%	0.6%	0.8%	0.6%	246
music_likeness	0.3%	0.4%	0.2%	0.3%	145
music_query	0.4%	0.5%	0.8%	0.5%	220
music_settings	0.1%	0.1%	0.2%	0.1%	68
news_query	1.4%	0.9%	1.0%	1.3%	749
newsletter_subscription	1.6%	0.7%	1.8%	1.7%	861
payment_issue	1.5%	1.6%	1.7%	2.3%	829
place_order	1.6%	2.5%	1.2%	1.5%	870
play_audiobook	0.3%	0.1%	0.3%	0.4%	172
play_game	0.3%	0.6%	0.3%	0.4%	191
play_music	5.2%	5.5%	4.3%	3.9%	2,855
play_podcasts	0.5%	0.2%	0.7%	0.7%	295
play_radio	0.8%	0.8%	0.7%	0.9%	436
qa_currency	0.6%	0.5%	0.9%	0.8%	311
qa_definition	0.7%	0.3%	1.2%	0.8%	376
qa_factoid	1.5%	1.4%	1.6%	1.8%	815
qa_maths	0.2%	0.1%	0.2%	0.2%	122
qa_stock	0.4%	0.5%	0.3%	0.5%	216
rate_book	3.3%	3.2%	4.3%	1.8%	1,810
recommendation_events	0.5%	0.4%	0.7%	0.6%	271
recommendation_locations	0.4%	0.3%	0.3%	0.5%	210
recommendation_movies	0.2%	0.1%	0.3%	0.2%	90
recover_password	1.6%	1.5%	2.2%	1.6%	862
registration_problems	1.6%	2.1%	1.3%	1.8%	854
review	1.6%	1.5%	0.9%	1.7%	873
search_creative_work	3.5%	2.8%	2.5%	1.8%	1,897
search_screening_event	3.2%	3.1%	3.5%	1.8%	1,752
set_up_shipping_address	1.6%	1.4%	2.0%	1.5%	871
social_post	0.8%	0.8%	0.8%	1.0%	436
social_query	0.3%	0.2%	0.2%	0.3%	159
switch_account	1.3%	1.8%	1.1%	1.5%	686
takeaway_order	0.3%	0.4%	0.4%	0.4%	160
takeaway_query	0.3%	0.8%	0.2%	0.3%	182
track_order	1.2%	1.1%	1.2%	1.1%	672
track_refund	1.1%	1.4%	1.3%	1.1%	597
transport_query	0.6%	0.7%	0.7%	0.6%	330
transport_taxi	0.3%	0.2%	0.5%	0.4%	150
transport_ticket	0.3%	0.2%	0.3%	0.4%	191
transport_traffic	0.3%	0.2%	0.3%	0.2%	167
weather_query	1.6%	1.6%	1.6%	1.6%	867

Row kind share per split

kind	train	dev	calibration	test	train rows
intent	100.0%	100.0%	100.0%	100.0%	54,849

4. Format

split	row-level format problems
train	none
dev	none
calibration	none
test	none

Options per choice row

split	min	median	p99	max
train	7	27	64	64
dev	7	27	64	64
calibration	7	27	64	64
test	7	27	64	64

Prompt length in tokens

split	measure	median	p99	max
train	estimate: characters / 3 (upper bound for English)	629	1181	1224
dev	estimate: characters / 3 (upper bound for English)	629	1183	1199
calibration	estimate: characters / 3 (upper bound for English)	629	1182	1217
test	estimate: characters / 3 (upper bound for English)	632	1182	1204

State key sets (train)

keys	rows
message, service	54,849 (100.0%)

Instructions (train)

Canonical (the most common text) 70.0%, reworded 26.9% (20 distinct rewordings), none 3.1%. Target about 70 / 27 / 3.
Canonical text: "What does the customer want? Choose the request that best matches what the customer is asking for in their message."
source instruction tag: canonical 70.0%, none 3.1%, variant-3 1.4%, variant-8 1.4%, variant-6 1.4%, variant-5 1.4%

class	canonical	none
add_to_playlist	70.3%	3.1%
alarm_query	71.9%	1.9%
alarm_remove	60.4%	5.0%
alarm_set	74.8%	2.1%
audio_volume_down	65.2%	7.6%
audio_volume_mute	74.0%	3.1%
audio_volume_up	67.8%	1.7%
book_restaurant	70.6%	2.9%
calendar_query	71.6%	2.6%
calendar_remove	70.8%	2.1%
calendar_set	69.7%	3.1%
cancel_order	70.1%	1.5%
change_order	70.7%	2.3%
change_shipping_address	70.3%	2.6%
check_cancellation_fee	69.0%	2.6%
check_invoice	70.0%	3.0%
check_payment_methods	70.1%	3.3%
check_refund_policy	74.1%	2.5%
complaint	66.7%	2.5%
contact_customer_service	70.5%	3.2%
contact_human_agent	69.6%	2.8%
cooking_recipe	69.2%	2.9%
create_account	73.1%	3.6%
datetime_convert	66.2%	2.7%
datetime_query	69.2%	3.6%
delete_account	69.7%	2.2%
delivery_options	68.6%	3.0%
delivery_period	70.6%	3.4%
edit_account	70.9%	3.8%
email_addcontact	75.7%	4.1%
email_query	73.9%	3.2%
email_querycontact	70.8%	1.8%
email_sendemail	71.3%	2.7%
general_affirm	72.3%	3.4%
general_commandstop	69.8%	1.8%
general_confirm	71.2%	4.2%
general_dontcare	67.8%	5.3%
general_explain	71.2%	2.3%
general_joke	70.3%	4.0%
general_negate	68.3%	3.7%
general_praise	69.3%	2.5%
general_quirky	70.4%	2.5%
general_repeat	70.2%	3.8%
get_invoice	71.8%	2.7%
get_refund	73.2%	2.4%
get_weather	69.2%	4.4%
iot_cleaning	72.1%	3.1%
iot_coffee	64.6%	3.2%
iot_hue_lightchange	70.4%	4.8%
iot_hue_lightdim	74.1%	2.8%
iot_hue_lightoff	67.5%	4.7%
iot_hue_lighton	73.3%	0.0%
iot_hue_lightup	77.9%	2.7%
iot_wemo_off	78.0%	1.2%
iot_wemo_on	75.0%	0.0%
lists_createoradd	68.8%	3.4%
lists_query	66.3%	3.4%
lists_remove	70.3%	2.0%
music_likeness	69.0%	4.1%
music_query	69.5%	0.9%
music_settings	61.8%	7.4%
news_query	68.0%	3.2%
newsletter_subscription	69.9%	2.2%
payment_issue	66.7%	3.1%
place_order	67.9%	3.2%
play_audiobook	65.7%	2.9%
play_game	66.5%	3.1%
play_music	70.3%	3.2%
play_podcasts	70.5%	3.7%
play_radio	71.1%	1.6%
qa_currency	72.7%	2.9%
qa_definition	72.1%	4.0%
qa_factoid	69.6%	2.6%
qa_maths	69.7%	4.9%
qa_stock	71.3%	3.7%
rate_book	69.1%	3.1%
recommendation_events	68.6%	3.0%
recommendation_locations	68.1%	2.4%
recommendation_movies	61.1%	3.3%
recover_password	70.5%	3.1%
registration_problems	71.4%	2.2%
review	68.4%	2.3%
search_creative_work	69.5%	2.2%
search_screening_event	68.1%	3.7%
set_up_shipping_address	70.1%	2.5%
social_post	68.6%	5.0%
social_query	64.8%	5.0%
switch_account	68.7%	4.7%
takeaway_order	66.9%	3.1%
takeaway_query	70.3%	3.8%
track_order	69.8%	3.1%
track_refund	68.0%	5.5%
transport_query	71.2%	3.6%
transport_taxi	70.7%	6.0%
transport_ticket	65.4%	3.7%
transport_traffic	71.3%	1.8%
weather_query	72.5%	3.8%

1. Shortcuts

Phrase statistics and models use a label-stratified sample of 50,000 train rows; models are scored on the full test file (5,577 rows).

Text length by label class (main text, characters)

split	class	rows	p10	median	p90	mean
train	add_to_playlist	1863	32	47	64	48
train	alarm_query	160	20	32	46	32
train	alarm_remove	101	17	26	45	29
train	alarm_set	242	18	31	54	35
train	audio_volume_down	66	12	23	35	24
train	audio_volume_mute	131	10	21	36	21
train	audio_volume_up	121	12	24	44	26
train	book_restaurant	1904	36	57	85	59
train	calendar_query	842	23	35	55	37
train	calendar_remove	435	21	35	53	36
train	calendar_set	1186	26	46	71	47
train	cancel_order	388	33	44	55	43
train	change_order	744	31	48	57	46
train	change_shipping_address	833	33	45	55	44
train	check_cancellation_fee	815	34	45	55	44
train	check_invoice	791	27	47	59	44
train	check_payment_methods	854	37	48	58	47
train	check_refund_policy	829	40	54	59	51
train	complaint	867	33	51	59	48
train	contact_customer_service	870	36	53	59	50
train	contact_human_agent	833	26	37	53	37
train	cooking_recipe	315	19	32	49	33
train	create_account	743	24	44	58	41
train	datetime_convert	74	30	54	66	50
train	datetime_query	504	17	30	44	30
train	delete_account	776	24	50	59	46
train	delivery_options	558	35	50	58	47
train	delivery_period	853	43	52	59	51
train	edit_account	653	25	34	54	36
train	email_addcontact	74	19	37	62	38
train	email_query	625	18	32	51	33
train	email_querycontact	171	18	31	51	32
train	email_sendemail	554	18	37	71	41
train	general_affirm	469	13	25	42	26
train	general_commandstop	275	17	25	32	24
train	general_confirm	480	25	36	49	36
train	general_dontcare	379	18	28	37	27
train	general_explain	598	26	38	55	39
train	general_joke	101	11	21	37	23
train	general_negate	807	22	36	49	35
train	general_praise	683	20	31	41	30
train	general_quirky	866	13	29	60	33
train	general_repeat	500	18	33	50	33
train	get_invoice	753	29	48	58	45
train	get_refund	781	29	43	56	42
train	get_weather	1919	29	47	70	48
train	iot_cleaning	129	15	26	43	28
train	iot_coffee	158	15	24	45	28
train	iot_hue_lightchange	186	19	32	48	33
train	iot_hue_lightdim	108	17	27	39	27
train	iot_hue_lightoff	191	17	27	40	28
train	iot_hue_lighton	30	14	22	36	23
train	iot_hue_lightup	113	15	27	42	28
train	iot_wemo_off	82	16	27	39	28
train	iot_wemo_on	68	13	23	36	24
train	lists_createoradd	237	15	29	45	30
train	lists_query	297	15	27	44	28
train	lists_remove	246	16	28	46	30
train	music_likeness	145	16	29	54	32
train	music_query	220	18	31	48	32
train	music_settings	68	13	27	46	29
train	news_query	749	16	34	54	35
train	newsletter_subscription	861	39	51	59	49
train	payment_issue	829	36	48	58	47
train	place_order	870	28	39	52	39
train	play_audiobook	172	16	32	50	33
train	play_game	191	14	24	40	24
train	play_music	2855	19	33	53	34
train	play_podcasts	295	17	30	48	32
train	play_radio	436	16	31	53	32
train	qa_currency	311	25	41	59	41
train	qa_definition	376	16	26	40	27
train	qa_factoid	815	21	32	50	34
train	qa_maths	122	21	32	49	34
train	qa_stock	216	20	33	52	35
train	rate_book	1810	26	38	61	41
train	recommendation_events	271	21	36	56	37
train	recommendation_locations	210	23	37	54	37
train	recommendation_movies	90	20	36	58	38
train	recover_password	862	32	45	56	44
train	registration_problems	854	32	45	57	44
train	review	873	38	49	57	47
train	search_creative_work	1897	25	39	57	40
train	search_screening_event	1752	32	49	67	49
train	set_up_shipping_address	871	38	48	57	47
train	social_post	436	24	42	70	44
train	social_query	159	22	37	60	39
train	switch_account	686	27	42	56	41
train	takeaway_order	160	18	36	61	38
train	takeaway_query	182	21	32	50	34
train	track_order	672	33	47	57	46
train	track_refund	597	35	50	59	48
train	transport_query	330	23	42	62	42
train	transport_taxi	150	13	29	57	32
train	transport_ticket	191	21	44	67	44
train	transport_traffic	167	21	36	52	36
train	weather_query	867	19	33	49	33
test	add_to_playlist	100	38	49	69	52
test	alarm_query	21	13	28	38	27
test	alarm_remove	13	16	23	41	27
test	alarm_set	30	17	30	48	32
test	audio_volume_down	9	18	23	30	23
test	audio_volume_mute	15	16	23	36	24
test	audio_volume_up	3	25	27	54	37
test	book_restaurant	100	35	56	82	58
test	calendar_query	84	24	35	53	37
test	calendar_remove	56	21	30	55	35
test	calendar_set	150	31	47	69	49
test	cancel_order	26	33	46	56	43
test	change_order	75	27	47	55	43
test	change_shipping_address	97	34	45	54	44
test	check_cancellation_fee	90	29	43	54	42
test	check_invoice	88	30	50	60	47
test	check_payment_methods	104	37	49	57	48
test	check_refund_policy	124	40	55	59	52
test	complaint	106	33	50	59	47
test	contact_customer_service	85	36	52	59	49
test	contact_human_agent	93	26	38	57	39
test	cooking_recipe	47	16	32	51	33
test	create_account	82	25	44	56	41
test	datetime_convert	4	35	54	55	47
test	datetime_query	63	15	28	41	28
test	delete_account	89	23	52	60	47
test	delivery_options	62	33	51	57	47
test	delivery_period	103	42	52	59	51
test	edit_account	78	26	35	55	37
test	email_addcontact	12	22	38	53	37
test	email_query	69	20	32	50	33
test	email_querycontact	23	18	34	48	32
test	email_sendemail	75	18	34	73	39
test	general_affirm	69	15	27	44	28
test	general_commandstop	38	17	24	31	24
test	general_confirm	53	23	37	48	35
test	general_dontcare	58	19	28	41	29
test	general_explain	64	27	40	56	40
test	general_joke	12	18	27	38	27
test	general_negate	104	24	36	48	35
test	general_praise	76	21	32	42	31
test	general_quirky	115	10	30	62	33
test	general_repeat	63	18	37	51	34
test	get_invoice	94	34	49	60	47
test	get_refund	84	27	43	58	43
test	get_weather	99	30	50	77	52
test	iot_cleaning	23	17	24	48	30
test	iot_coffee	20	15	26	43	27
test	iot_hue_lightchange	18	19	32	52	34
test	iot_hue_lightdim	11	17	25	32	24
test	iot_hue_lightoff	26	19	31	49	33
test	iot_hue_lighton	7	14	24	47	28
test	iot_hue_lightup	17	17	26	49	32
test	iot_wemo_off	8	14	30	39	27
test	iot_wemo_on	6	17	24	29	23
test	lists_createoradd	27	12	27	44	28
test	lists_query	30	18	24	41	27
test	lists_remove	35	17	29	47	31
test	music_likeness	19	25	31	51	36
test	music_query	30	16	29	51	33
test	music_settings	5	14	21	60	33
test	news_query	71	17	33	57	34
test	newsletter_subscription	96	39	51	58	50
test	payment_issue	130	37	46	58	46
test	place_order	82	28	40	52	40
test	play_audiobook	24	17	31	50	33
test	play_game	22	15	26	44	28
test	play_music	220	17	30	51	33
test	play_podcasts	41	17	30	44	31
test	play_radio	49	16	32	47	33
test	qa_currency	44	20	37	53	36
test	qa_definition	47	14	26	41	28
test	qa_factoid	99	20	33	49	33
test	qa_maths	13	23	32	84	45
test	qa_stock	30	27	35	46	36
test	rate_book	98	26	36	51	37
test	recommendation_events	31	21	37	51	37
test	recommendation_locations	30	22	42	57	40
test	recommendation_movies	9	18	30	64	37
test	recover_password	88	31	44	55	43
test	registration_problems	101	34	46	57	45
test	review	95	39	49	57	48
test	search_creative_work	100	32	42	60	44
test	search_screening_event	98	29	46	62	45
test	set_up_shipping_address	84	37	48	57	47
test	social_post	53	20	44	72	47
test	social_query	14	15	39	48	38
test	switch_account	81	25	43	54	40
test	takeaway_order	20	17	46	61	42
test	takeaway_query	16	33	41	56	42
test	track_order	64	35	50	57	48
test	track_refund	60	33	51	59	48
test	transport_query	35	22	43	67	45
test	transport_taxi	21	14	27	51	30
test	transport_ticket	25	28	46	77	50
test	transport_traffic	11	18	29	46	32
test	weather_query	88	21	32	49	34

By row kind (train): main-text length, length of the rest of the state, options

kind	rows	median chars	mean chars	median other-state chars	median options
intent	54849	41	41	40	27

Correct option: longest / shortest / position / key

For rows whose answer is one of the listed options (fixed options such as 'none of these' excluded):

split	rows	correct is longest	correct is shortest	chance (1/listed)	mean relative position (0 first, 1 last; 0.5 expected)	position fifths

Correct key and position by option count

split	options	rows	mean options	top correct keys	most common position (0-based)
train	6-10	13036	7.0	get_weather 14.7%, book_restaurant 14.6%, search_creative_work 14.6%, play_music 14.5%, add_to_playlist 14.3%	0 (14.8%)
train	11-30	20916	27.0	review 4.2%, set_up_shipping_address 4.2%, place_order 4.2%, contact_customer_service 4.2%, complaint 4.1%	12 (4.0%)
train	31-80	20897	64.0	calendar_set 5.7%, play_music 4.6%, weather_query 4.1%, general_quirky 4.1%, calendar_query 4.0%	52 (1.8%)
test	6-10	695	7.0	search_creative_work 14.4%, add_to_playlist 14.4%, play_music 14.4%, book_restaurant 14.4%, get_weather 14.2%	4 (17.3%)
test	11-30	2361	27.0	payment_issue 5.5%, check_refund_policy 5.3%, complaint 4.5%, check_payment_methods 4.4%, delivery_period 4.4%	14 (4.2%)
test	31-80	2521	64.0	calendar_set 6.0%, play_music 4.8%, general_quirky 4.6%, general_negate 4.1%, qa_factoid 3.9%	63 (2.3%)

Option count by label class (train)

class	rows	min	median	mean	max
add_to_playlist	1863	7	7	7.0	7
alarm_query	160	64	64	64.0	64
alarm_remove	101	64	64	64.0	64
alarm_set	242	64	64	64.0	64
audio_volume_down	66	64	64	64.0	64
audio_volume_mute	131	64	64	64.0	64
audio_volume_up	121	64	64	64.0	64
book_restaurant	1904	7	7	7.0	7
calendar_query	842	64	64	64.0	64
calendar_remove	435	64	64	64.0	64
calendar_set	1186	64	64	64.0	64
cancel_order	388	27	27	27.0	27
change_order	744	27	27	27.0	27
change_shipping_address	833	27	27	27.0	27
check_cancellation_fee	815	27	27	27.0	27
check_invoice	791	27	27	27.0	27
check_payment_methods	854	27	27	27.0	27
check_refund_policy	829	27	27	27.0	27
complaint	867	27	27	27.0	27
contact_customer_service	870	27	27	27.0	27
contact_human_agent	833	27	27	27.0	27
cooking_recipe	315	64	64	64.0	64
create_account	743	27	27	27.0	27
datetime_convert	74	64	64	64.0	64
datetime_query	504	64	64	64.0	64
delete_account	776	27	27	27.0	27
delivery_options	558	27	27	27.0	27
delivery_period	853	27	27	27.0	27
edit_account	653	27	27	27.0	27
email_addcontact	74	64	64	64.0	64
email_query	625	64	64	64.0	64
email_querycontact	171	64	64	64.0	64
email_sendemail	554	64	64	64.0	64
general_affirm	469	64	64	64.0	64
general_commandstop	275	64	64	64.0	64
general_confirm	480	64	64	64.0	64
general_dontcare	379	64	64	64.0	64
general_explain	598	64	64	64.0	64
general_joke	101	64	64	64.0	64
general_negate	807	64	64	64.0	64
general_praise	683	64	64	64.0	64
general_quirky	866	64	64	64.0	64
general_repeat	500	64	64	64.0	64
get_invoice	753	27	27	27.0	27
get_refund	781	27	27	27.0	27
get_weather	1919	7	7	7.0	7
iot_cleaning	129	64	64	64.0	64
iot_coffee	158	64	64	64.0	64
iot_hue_lightchange	186	64	64	64.0	64
iot_hue_lightdim	108	64	64	64.0	64
iot_hue_lightoff	191	64	64	64.0	64
iot_hue_lighton	30	64	64	64.0	64
iot_hue_lightup	113	64	64	64.0	64
iot_wemo_off	82	64	64	64.0	64
iot_wemo_on	68	64	64	64.0	64
lists_createoradd	237	64	64	64.0	64
lists_query	297	64	64	64.0	64
lists_remove	246	64	64	64.0	64
music_likeness	145	64	64	64.0	64
music_query	220	64	64	64.0	64
music_settings	68	64	64	64.0	64
news_query	749	64	64	64.0	64
newsletter_subscription	861	27	27	27.0	27
payment_issue	829	27	27	27.0	27
place_order	870	27	27	27.0	27
play_audiobook	172	64	64	64.0	64
play_game	191	64	64	64.0	64
play_music	2855	7	7	26.2	64
play_podcasts	295	64	64	64.0	64
play_radio	436	64	64	64.0	64
qa_currency	311	64	64	64.0	64
qa_definition	376	64	64	64.0	64
qa_factoid	815	64	64	64.0	64
qa_maths	122	64	64	64.0	64
qa_stock	216	64	64	64.0	64
rate_book	1810	7	7	7.0	7
recommendation_events	271	64	64	64.0	64
recommendation_locations	210	64	64	64.0	64
recommendation_movies	90	64	64	64.0	64
recover_password	862	27	27	27.0	27
registration_problems	854	27	27	27.0	27
review	873	27	27	27.0	27
search_creative_work	1897	7	7	7.0	7
search_screening_event	1752	7	7	7.0	7
set_up_shipping_address	871	27	27	27.0	27
social_post	436	64	64	64.0	64
social_query	159	64	64	64.0	64
switch_account	686	27	27	27.0	27
takeaway_order	160	64	64	64.0	64
takeaway_query	182	64	64	64.0	64
track_order	672	27	27	27.0	27
track_refund	597	27	27	27.0	27
transport_query	330	64	64	64.0	64
transport_taxi	150	64	64	64.0	64
transport_ticket	191	64	64	64.0	64
transport_traffic	167	64	64	64.0	64
weather_query	867	64	64	64.0	64

Source fields by label class (train)

Scalar source fields with 2–60 values. 'Purity' = accuracy of predicting the label class from this field alone (per-value majority), against the overall majority. The model does not see source, but a field that predicts the label marks a confound: rows of one origin carry one label, so any style difference of that origin becomes a shortcut.

Overall majority: 5.2%.

source field	values	purity	top values → classes
dataset	3	7.3%	bitext: review 4.2%, set_up_shipping_address 4.2%; hwu64: calendar_set 5.7%, play_music 4.6%; snips: get_weather 14.7%, book_restaurant 14.6%
url	3	7.3%	https://huggingface.co/datasets/bitext/Bitext-customer-support-llm-chatbot-training-dataset: review 4.2%, set_up_shipping_address 4.2%; https://github.com/xliuhw/NLU-Evaluation-Data: calendar_set 5.7%, play_music 4.6%; https://github.com/sonos/nlu-benchmark: get_weather 14.7%, book_restaurant 14.6%
revision	3	7.3%	430d1a89bd93bd1fa23c16f29dd53e73f0087443: review 4.2%, set_up_shipping_address 4.2%; f6071b496b17d71e6eb43f543af0707f4ff30557: calendar_set 5.7%, play_music 4.6%; b86ac7f1577868c42158d0dec77db50956046696: get_weather 14.7%, book_restaurant 14.6%
license	3	7.3%	CDLA-Sharing-1.0 (models trained on the data are Results, not Data, under CDLA): review 4.2%, set_up_shipping_address 4.2%; CC BY 4.0 (repository LICENSE and README): calendar_set 5.7%, play_music 4.6%; CC0 1.0 (repository licence): get_weather 14.7%, book_restaurant 14.6%
options_from	3	7.3%	bitext: review 4.2%, set_up_shipping_address 4.2%; hwu64: calendar_set 5.7%, play_music 4.6%; snips: get_weather 14.7%, book_restaurant 14.6%
split	2	5.7%	null: calendar_set 2.8%, play_music 2.3%; train: get_weather 14.7%, book_restaurant 14.6%
copies	15	5.7%	1: play_music 5.3%, get_weather 3.6%; 2: track_order 10.5%, switch_account 7.8%; 3: edit_account 30.0%, switch_account 14.1%; 5: cancel_order 31.5%, track_refund 24.5%; 4: cancel_order 34.5%, track_order 12.1%; 6: cancel_order 58.6%, delivery_options 31.0%
target_kind	2	5.2%	hard: play_music 5.2%, get_weather 3.5%; soft: general_praise 33.3%, general_affirm 33.3%
instructions	22	5.2%	canonical: play_music 5.2%, book_restaurant 3.5%; none: play_music 5.4%, get_weather 4.9%; variant-3: play_music 5.6%, search_screening_event 4.6%; variant-8: play_music 4.6%, book_restaurant 4.3%; variant-6: play_music 6.3%, book_restaurant 4.9%; variant-5: rate_book 4.3%, play_music 3.9%

Formatting by label class (main text, share of rows)

feature	lowest classes	highest classes
ends with ?	alarm_query 0%, alarm_remove 0%, alarm_set 0%	delivery_options 32%, contact_customer_service 30%, get_weather 27%	gap
ends with .	alarm_query 0%, alarm_remove 0%, alarm_set 0%	general_repeat 100%, general_praise 100%, general_negate 100%	gap
ends with !	add_to_playlist 0%, alarm_query 0%, alarm_remove 0%	search_creative_work 0%, play_music 0%, weather_query 0%
no end punctuation	general_affirm 0%, general_commandstop 0%, general_confirm 0%	weather_query 100%, transport_traffic 100%, transport_ticket 100%	gap
starts lowercase	get_weather 19%, search_screening_event 26%, search_creative_work 28%	transport_traffic 100%, transport_taxi 100%, takeaway_query 100%	gap
all lowercase	search_creative_work 0%, get_weather 2%, book_restaurant 7%	transport_taxi 100%, takeaway_query 100%, takeaway_order 100%	gap
has a digit	alarm_query 0%, alarm_remove 0%, alarm_set 0%	change_order 99%, track_order 98%, cancel_order 97%	gap
has newline	alarm_query 0%, alarm_remove 0%, alarm_set 0%	rate_book 1%, search_screening_event 0%, add_to_playlist 0%
has quotes	alarm_query 0%, alarm_remove 0%, alarm_set 0%	add_to_playlist 0%, play_music 0%, weather_query 0%
has markup (HTML/markdown)	add_to_playlist 0%, alarm_query 0%, alarm_remove 0%	weather_query 0%, transport_traffic 0%, transport_ticket 0%
has URL	add_to_playlist 0%, alarm_query 0%, alarm_remove 0%	weather_query 0%, transport_traffic 0%, transport_ticket 0%
non-ASCII	alarm_query 0%, alarm_remove 0%, alarm_set 0%	add_to_playlist 7%, track_refund 7%, get_refund 7%
non-Latin script	add_to_playlist 0%, alarm_query 0%, alarm_remove 0%	search_creative_work 0%, weather_query 0%, transport_traffic 0%
emoji	add_to_playlist 0%, alarm_query 0%, alarm_remove 0%	play_music 0%, weather_query 0%, transport_traffic 0%
ALL-CAPS word (4+)	alarm_query 0%, alarm_remove 0%, alarm_set 0%	add_to_playlist 1%, search_screening_event 1%, search_creative_work 1%
contains ' - ' or —	add_to_playlist 0%, alarm_query 0%, alarm_remove 0%	search_creative_work 1%, search_screening_event 0%, book_restaurant 0%

Same, by row kind

feature	intent
ends with ?	10%
ends with .	12%
ends with !	0%
no end punctuation	79%
starts lowercase	70%
all lowercase	59%
has a digit	12%
has newline	0%
has quotes	0%
has markup (HTML/markdown)	0%
has URL	0%
non-ASCII	1%
non-Latin script	0%
emoji	0%
ALL-CAPS word (4+)	0%
contains ' - ' or —	0%

Over-represented words and phrases per label class (main text)

Log-odds ratio with an informative Dirichlet prior (Monroe et al. 2008), each class against all the others; z-score, then the share of rows in the class and in the other classes that contain the phrase. Counted once per row.

Words, add_to_playlist: add z=82 81.1% vs 0.9%; playlist z=69 63.6% vs 0.5%; my z=39 52.3% vs 12.8%; to z=36 81.0% vs 30.3%; this z=34 23.7% vs 3.5%; put z=33 12.2% vs 0.3%; tune z=29 10.4% vs 0.1%; track z=28 9.1% vs 0.4%; artist z=27 8.4% vs 0.1%; album z=24 8.2% vs 0.6%; song z=22 9.3% vs 1.2%; metal z=16 3.4% vs 0.0%; pop z=15 2.7% vs 0.1%; rock z=15 3.2% vs 0.2%; hits z=15 2.4% vs 0.1%; into z=15 2.4% vs 0.0%; de z=14 2.1% vs 0.1%; women z=14 2.1% vs 0.0%; dance z=14 2.0% vs 0.1%; indie z=13 2.9% vs 0.0%

Words, alarm_query: set z=28 56.8% vs 1.3%; alarm z=27 41.8% vs 0.5%; alarms z=27 57.5% vs 0.0%; morning z=14 13.7% vs 0.3%; tomorrow z=10 12.3% vs 0.7%; any z=9 14.4% vs 1.4%; have z=8 30.1% vs 6.2%; are z=7 15.8% vs 2.6%; wake z=7 3.4% vs 0.1%; time z=7 13.7% vs 2.1%; active z=6 2.1% vs 0.0%; let z=6 4.8% vs 0.3%; currently z=6 2.7% vs 0.1%; today z=5 9.6% vs 1.5%; for z=5 30.8% vs 11.4%; many z=5 4.8% vs 0.4%; checkout z=5 1.4% vs 0.0%; i've z=5 2.1% vs 0.1%; settings z=5 1.4% vs 0.0%; what z=5 28.8% vs 11.7%

Words, alarm_remove: alarm z=30 69.6% vs 0.5%; remove z=16 31.5% vs 0.7%; alarms z=16 20.7% vs 0.2%; disable z=9 6.5% vs 0.0%; wake z=8 6.5% vs 0.1%; set z=8 18.5% vs 1.5%; rid z=8 4.3% vs 0.0%; am z=7 19.6% vs 2.0%; off z=7 12.0% vs 0.7%; delete z=7 10.9% vs 0.6%; removed z=7 3.3% vs 0.0%; calls z=7 3.3% vs 0.0%; cancel z=7 13.0% vs 1.0%; pm z=7 12.0% vs 0.9%; eight z=6 7.6% vs 0.3%; wednesday z=6 4.3% vs 0.1%; tomorrow z=6 9.8% vs 0.8%; all z=5 9.8% vs 0.9%; turn z=5 8.7% vs 0.8%; longer z=4 2.2% vs 0.1%

Words, alarm_set: alarm z=46 77.8% vs 0.3%; set z=36 59.3% vs 1.2%; am z=25 41.6% vs 1.8%; wake z=18 14.5% vs 0.0%; six z=18 14.9% vs 0.3%; an z=18 35.3% vs 3.3%; at z=15 38.5% vs 5.3%; tomorrow z=15 15.8% vs 0.7%; morning z=15 11.3% vs 0.3%; five z=15 16.3% vs 0.8%; thirty z=14 9.0% vs 0.2%; pm z=14 15.4% vs 0.9%; seven z=13 10.9% vs 0.4%; for z=12 47.5% vs 11.3%; eight z=12 8.6% vs 0.3%; up z=10 17.6% vs 2.6%; alert z=10 3.6% vs 0.0%; clock z=9 3.2% vs 0.0%; ten z=8 6.3% vs 0.5%; before z=8 4.1% vs 0.2%

Words, audio_volume_down: volume z=24 60.0% vs 0.2%; lower z=18 33.3% vs 0.1%; down z=14 26.7% vs 0.2%; speakers z=10 11.7% vs 0.0%; turn z=10 26.7% vs 0.8%; percent z=10 10.0% vs 0.0%; decrease z=9 11.7% vs 0.0%; speaker z=9 8.3% vs 0.0%; low z=8 6.7% vs 0.0%; loud z=7 5.0% vs 0.0%; too z=5 5.0% vs 0.1%; please z=5 28.3% vs 5.3%; player z=4 3.3% vs 0.1%; audio z=4 3.3% vs 0.1%; adjust z=4 1.7% vs 0.0%; little z=4 3.3% vs 0.1%; bass z=4 1.7% vs 0.0%; reduce z=4 1.7% vs 0.0%; you're z=3 1.7% vs 0.0%; sound z=3 3.3% vs 0.1%

Words, audio_volume_mute: speakers z=15 12.6% vs 0.0%; sound z=13 10.1% vs 0.1%; until z=12 7.6% vs 0.0%; quiet z=11 12.6% vs 0.0%; mode z=10 5.9% vs 0.0%; off z=10 14.3% vs 0.7%; speaker z=10 5.9% vs 0.0%; volume z=10 9.2% vs 0.2%; mute z=10 37.8% vs 0.0%; silence z=8 4.2% vs 0.0%; please z=8 26.1% vs 5.3%; turn z=7 10.1% vs 0.8%; minutes z=7 5.9% vs 0.2%; pause z=7 2.5% vs 0.0%; yourself z=7 2.5% vs 0.0%; noise z=7 2.5% vs 0.0%; stop z=6 4.2% vs 0.2%; control z=6 1.7% vs 0.0%; audio z=5 2.5% vs 0.1%; settings z=5 1.7% vs 0.0%

Words, audio_volume_up: volume z=32 63.6% vs 0.1%; increase z=18 19.1% vs 0.0%; turn z=15 24.5% vs 0.8%; speaker z=12 8.2% vs 0.0%; raise z=12 8.2% vs 0.0%; up z=11 27.3% vs 2.6%; percent z=10 6.4% vs 0.0%; player z=10 6.4% vs 0.0%; speakers z=10 6.4% vs 0.0%; fifty z=8 5.5% vs 0.1%; louder z=8 13.6% vs 0.0%; sound z=8 5.5% vs 0.1%; level z=7 3.6% vs 0.0%; music z=7 14.5% vs 1.7%; speak z=7 9.1% vs 0.7%; unmute z=7 4.5% vs 0.0%; please z=6 23.6% vs 5.3%; sixty z=6 2.7% vs 0.0%; little z=6 3.6% vs 0.1%; loud z=5 1.8% vs 0.0%

Words, book_restaurant: book z=70 61.9% vs 1.1%; at z=52 48.2% vs 3.9%; for z=52 69.9% vs 9.4%; in z=49 61.5% vs 8.1%; a z=48 86.0% vs 16.0%; restaurant z=43 37.2% vs 0.2%; people z=27 11.5% vs 0.1%; and z=26 19.8% vs 2.8%; party z=23 7.5% vs 0.3%; that z=23 14.2% vs 1.9%; table z=21 26.3% vs 0.0%; spot z=20 6.9% vs 0.0%; rated z=20 6.7% vs 0.1%; food z=20 5.3% vs 0.2%; eat z=19 5.5% vs 0.0%; bar z=17 5.9% vs 0.0%; reservation z=17 14.1% vs 0.0%; 8 z=17 3.6% vs 0.1%; 10 z=16 3.2% vs 0.1%; within z=15 3.1% vs 0.1%

Words, calendar_query: today z=25 16.4% vs 1.3%; events z=25 11.2% vs 0.4%; meeting z=24 12.5% vs 0.8%; scheduled z=19 5.9% vs 0.1%; calendar z=19 9.9% vs 0.9%; reminders z=19 8.1% vs 0.0%; appointment z=18 5.5% vs 0.2%; meetings z=18 6.0% vs 0.0%; have z=18 23.8% vs 6.0%; next z=17 9.2% vs 1.0%; week z=17 7.2% vs 0.6%; is z=16 28.3% vs 9.2%; tomorrow z=16 7.3% vs 0.7%; any z=16 9.5% vs 1.3%; event z=15 6.8% vs 0.7%; upcoming z=14 3.3% vs 0.1%; schedule z=14 6.5% vs 0.7%; do z=14 21.9% vs 7.4%; appointments z=13 2.9% vs 0.0%; what z=13 28.6% vs 11.5%

Words, calendar_remove: calendar z=47 48.4% vs 0.6%; delete z=37 31.5% vs 0.4%; remove z=33 25.9% vs 0.5%; all z=32 27.2% vs 0.7%; events z=31 22.9% vs 0.4%; event z=30 22.7% vs 0.6%; clear z=23 12.6% vs 0.2%; next z=18 13.4% vs 1.0%; appointment z=16 6.8% vs 0.2%; my z=15 43.1% vs 13.9%; erase z=15 6.0% vs 0.0%; cancel z=14 10.1% vs 1.0%; from z=14 21.9% vs 5.0%; scheduled z=13 4.5% vs 0.1%; meeting z=11 7.6% vs 0.9%; appointments z=11 2.8% vs 0.1%; tomorrow z=10 6.3% vs 0.7%; everything z=10 2.3% vs 0.0%; march z=10 3.5% vs 0.2%; birthday z=9 3.8% vs 0.3%

Words, calendar_set: meeting z=44 29.4% vs 0.3%; calendar z=38 21.1% vs 0.6%; event z=35 18.0% vs 0.4%; set z=32 19.3% vs 1.1%; pm z=31 15.5% vs 0.6%; remind z=28 23.2% vs 0.0%; reminder z=24 13.0% vs 0.0%; tomorrow z=23 10.1% vs 0.6%; at z=22 24.6% vs 5.0%; birthday z=21 6.4% vs 0.2%; every z=20 7.0% vs 0.0%; before z=18 4.8% vs 0.1%; on z=18 24.1% vs 6.3%; tuesday z=18 4.9% vs 0.1%; march z=17 4.6% vs 0.1%; friday z=17 4.5% vs 0.1%; add z=17 15.7% vs 3.3%; appointment z=16 4.3% vs 0.2%; two z=16 7.5% vs 0.8%; monday z=16 3.9% vs 0.1%

Words, cancel_order: cancel z=45 49.2% vs 0.7%; order z=38 50.3% vs 2.2%; purchase z=36 41.8% vs 1.5%; ord z=31 26.0% vs 0.6%; cancelling z=28 19.5% vs 0.1%; canceling z=27 18.4% vs 0.1%; bought z=14 5.1% vs 0.0%; longer z=12 3.7% vs 0.0%; i z=12 61.6% vs 32.1%; pay z=11 3.4% vs 0.1%; help z=10 26.0% vs 9.6%; can't z=9 4.2% vs 0.3%; item z=8 5.6% vs 0.8%; assistance z=8 13.8% vs 4.4%; need z=8 18.4% vs 7.4%; ordered z=8 1.4% vs 0.0%; question z=8 3.1% vs 0.3%; no z=7 3.7% vs 0.4%; with z=7 13.6% vs 5.4%; want z=7 15.0% vs 6.5%

Words, change_order: order z=54 54.3% vs 1.8%; purchase z=48 41.2% vs 1.3%; ord z=38 24.8% vs 0.5%; items z=25 10.6% vs 0.2%; item z=23 12.2% vs 0.7%; several z=23 10.8% vs 0.5%; something z=22 9.4% vs 0.3%; some z=22 18.0% vs 2.0%; articles z=21 7.8% vs 0.2%; products z=21 9.1% vs 0.4%; change z=18 10.2% vs 0.9%; help z=18 32.2% vs 9.4%; an z=18 17.7% vs 3.2%; switch z=17 6.5% vs 0.3%; article z=17 7.8% vs 0.6%; product z=17 9.0% vs 0.9%; edit z=16 6.2% vs 0.4%; remove z=16 7.5% vs 0.7%; delete z=16 7.1% vs 0.6%; from z=14 18.6% vs 5.0%

Words, change_shipping_address: address z=79 96.4% vs 1.8%; shipping z=40 29.9% vs 1.2%; delivery z=37 27.8% vs 1.4%; trying z=34 20.4% vs 0.6%; update z=29 14.5% vs 0.4%; modify z=26 11.5% vs 0.2%; edit z=24 9.9% vs 0.3%; correct z=23 10.7% vs 0.5%; change z=22 12.6% vs 0.9%; modifying z=22 8.3% vs 0.1%; editing z=21 7.0% vs 0.1%; correcting z=20 6.7% vs 0.1%; my z=20 44.8% vs 13.7%; updating z=20 6.3% vs 0.1%; changing z=18 5.5% vs 0.2%; about z=16 15.5% vs 3.0%; wrong z=15 5.0% vs 0.3%; information z=13 7.4% vs 1.0%; changed z=13 3.7% vs 0.0%; i z=12 58.8% vs 31.9%

Words, check_cancellation_fee: termination z=35 36.6% vs 0.1%; see z=32 35.9% vs 3.9%; check z=30 32.7% vs 3.7%; cancellation z=29 17.8% vs 0.1%; the z=28 93.1% vs 29.4%; early z=24 40.1% vs 0.0%; seeing z=18 9.3% vs 0.7%; charge z=17 18.0% vs 0.0%; checking z=17 8.1% vs 0.6%; charges z=17 14.9% vs 0.0%; exit z=14 21.5% vs 0.0%; i z=13 62.4% vs 31.8%; help z=12 25.8% vs 9.5%; fees z=12 16.2% vs 0.0%; penalty z=10 17.9% vs 0.0%; know z=10 13.7% vs 4.4%; do z=9 18.0% vs 7.4%; assistance z=9 12.4% vs 4.3%; cannot z=8 3.2% vs 0.5%; how z=8 17.8% vs 7.8%

Words, check_invoice: bill z=48 41.2% vs 0.6%; invoice z=44 34.4% vs 0.6%; look z=43 32.9% vs 0.4%; from z=31 40.1% vs 4.7%; 00108 z=29 15.1% vs 0.2%; 12588 z=29 15.0% vs 0.2%; 37777 z=28 14.4% vs 0.2%; 85632 z=27 13.3% vs 0.2%; locate z=24 10.4% vs 0.1%; bills z=24 10.3% vs 0.1%; invoices z=23 9.6% vs 0.1%; take z=16 7.2% vs 0.5%; at z=15 22.2% vs 5.2%; quick z=14 22.1% vs 0.0%; giving z=14 3.6% vs 0.0%; my z=13 37.7% vs 13.8%; know z=11 15.8% vs 4.4%; do z=11 21.8% vs 7.4%; give z=11 9.2% vs 1.9%; taking z=10 1.9% vs 0.0%

Words, check_payment_methods: payment z=77 96.1% vs 1.0%; options z=45 33.8% vs 0.4%; methods z=44 31.7% vs 0.5%; available z=37 22.0% vs 0.3%; your z=27 25.9% vs 3.0%; see z=26 29.4% vs 4.0%; list z=26 20.2% vs 1.7%; ur z=22 14.8% vs 1.3%; check z=20 22.2% vs 3.8%; help z=16 30.9% vs 9.4%; seeing z=15 7.6% vs 0.7%; checking z=14 6.3% vs 0.7%; are z=13 11.6% vs 2.5%; assistance z=12 15.5% vs 4.3%; accepted z=11 24.8% vs 0.0%; accept z=9 4.2% vs 0.0%; want z=9 16.2% vs 6.4%; i z=9 53.5% vs 31.9%; ya z=8 3.3% vs 0.5%; listing z=8 5.6% vs 0.0%

Words, check_refund_policy: ask z=42 34.7% vs 0.3%; which z=42 32.1% vs 0.3%; request z=37 28.3% vs 0.2%; can z=36 68.9% vs 9.5%; refund z=34 21.2% vs 0.4%; see z=33 41.0% vs 3.8%; refunds z=31 20.9% vs 0.1%; in z=31 58.7% vs 9.3%; back z=30 17.2% vs 0.2%; money z=26 17.6% vs 0.9%; reimbursement z=25 12.4% vs 0.3%; check z=25 30.3% vs 3.7%; reimbursements z=24 11.4% vs 0.1%; take z=22 10.4% vs 0.4%; seeing z=20 11.5% vs 0.7%; long z=19 10.3% vs 0.6%; i z=18 86.1% vs 31.4%; help z=17 37.6% vs 9.3%; under z=16 5.0% vs 0.1%; policy z=15 16.8% vs 0.0%

Words, complaint: complaint z=40 30.9% vs 0.2%; a z=39 90.8% vs 17.3%; make z=39 27.2% vs 0.8%; your z=38 38.2% vs 2.8%; company z=34 22.7% vs 0.8%; business z=34 20.1% vs 0.2%; customer z=33 25.9% vs 1.6%; ur z=32 22.3% vs 1.2%; consumer z=29 26.6% vs 0.1%; file z=24 26.6% vs 0.0%; against z=23 54.6% vs 0.0%; making z=17 5.7% vs 0.0%; help z=15 30.6% vs 9.4%; lodge z=13 26.2% vs 0.0%; experience z=11 2.4% vs 0.0%; happy z=10 2.0% vs 0.1%; organization z=9 12.3% vs 0.0%; claim z=9 40.0% vs 0.0%; to z=8 55.2% vs 31.6%; bad z=8 1.5% vs 0.1%

Words, contact_customer_service: customer z=69 95.1% vs 0.5%; service z=45 33.9% vs 0.5%; hours z=38 25.6% vs 0.3%; support z=37 38.0% vs 0.1%; call z=35 20.7% vs 0.3%; contact z=30 16.1% vs 0.4%; can z=27 49.9% vs 9.8%; talk z=27 13.9% vs 0.5%; what z=26 52.2% vs 11.1%; assistance z=25 29.3% vs 4.1%; time z=25 20.4% vs 1.9%; speak z=22 10.8% vs 0.5%; available z=19 8.8% vs 0.5%; reach z=18 12.7% vs 0.0%; see z=17 20.7% vs 4.1%; touch z=16 6.4% vs 0.0%; i z=16 73.3% vs 31.6%; help z=14 30.3% vs 9.4%; number z=14 3.8% vs 0.1%; at z=14 19.9% vs 5.2%

Words, contact_human_agent: with z=40 48.5% vs 4.8%; talk z=39 24.9% vs 0.3%; speak z=39 24.6% vs 0.3%; an z=35 33.9% vs 3.0%; person z=28 13.7% vs 0.1%; contact z=28 13.7% vs 0.5%; live z=27 11.6% vs 0.2%; someone z=22 10.4% vs 0.0%; to z=18 69.3% vs 31.4%; human z=18 11.5% vs 0.0%; speaking z=17 6.1% vs 0.0%; talking z=17 5.5% vs 0.0%; understand z=16 4.2% vs 0.1%; i z=15 64.7% vs 31.8%; know z=15 17.9% vs 4.4%; chat z=15 13.6% vs 0.0%; helpful z=14 3.6% vs 0.1%; agent z=14 35.0% vs 0.0%; a z=13 39.8% vs 18.1%; assistant z=13 11.3% vs 0.0%

Words, cooking_recipe: recipe z=18 26.5% vs 0.0%; cook z=17 19.9% vs 0.0%; chicken z=16 7.7% vs 0.0%; how z=16 39.0% vs 7.8%; make z=12 11.8% vs 1.2%; meal z=10 3.1% vs 0.1%; steak z=10 2.8% vs 0.0%; rice z=10 3.1% vs 0.0%; pasta z=9 2.4% vs 0.0%; egg z=9 2.4% vs 0.0%; for z=9 29.3% vs 11.4%; cheese z=9 2.4% vs 0.0%; cookies z=9 2.8% vs 0.0%; turkey z=9 2.8% vs 0.0%; spaghetti z=9 3.8% vs 0.0%; best z=9 5.2% vs 0.5%; should z=8 4.2% vs 0.3%; prepare z=8 2.1% vs 0.0%; do z=8 21.6% vs 7.5%; cake z=8 1.7% vs 0.0%

Words, create_account: account z=65 86.7% vs 3.6%; create z=41 32.1% vs 0.2%; open z=39 26.9% vs 0.5%; standard z=29 17.6% vs 0.6%; a z=28 70.6% vs 17.7%; pro z=24 12.0% vs 0.4%; platinum z=24 12.0% vs 0.4%; creating z=22 12.6% vs 0.0%; premium z=22 9.9% vs 0.4%; opening z=21 16.5% vs 0.0%; gold z=21 9.9% vs 0.4%; freemium z=21 9.2% vs 0.3%; dad z=18 7.2% vs 0.0%; wife z=17 5.3% vs 0.1%; daughter z=16 6.9% vs 0.0%; yet z=13 3.1% vs 0.0%; silver z=13 3.4% vs 0.1%; elite z=13 3.1% vs 0.1%; student z=12 3.0% vs 0.1%; basic z=11 2.5% vs 0.1%

Words, datetime_convert: time z=20 86.6% vs 2.1%; convert z=15 23.9% vs 0.0%; zone z=15 22.4% vs 0.0%; difference z=13 19.4% vs 0.0%; between z=12 20.9% vs 0.2%; eastern z=11 11.9% vs 0.0%; pacific z=10 10.4% vs 0.1%; york z=9 14.9% vs 0.3%; g z=8 7.5% vs 0.0%; india z=8 7.5% vs 0.1%; t z=8 7.5% vs 0.1%; central z=7 6.0% vs 0.0%; it's z=7 11.9% vs 0.4%; here z=7 13.4% vs 0.5%; m z=7 7.5% vs 0.1%; change z=6 17.9% vs 1.1%; london z=6 6.0% vs 0.1%; thirty z=6 7.5% vs 0.2%; lagos z=6 3.0% vs 0.0%; manila z=6 3.0% vs 0.0%

Words, datetime_query: time z=50 59.3% vs 1.7%; date z=28 16.6% vs 0.2%; is z=27 53.6% vs 9.1%; day z=25 15.9% vs 0.5%; what z=24 53.4% vs 11.4%; in z=19 39.4% vs 9.7%; current z=15 13.5% vs 1.7%; tell z=15 13.7% vs 1.8%; now z=15 11.1% vs 1.2%; it z=14 22.4% vs 5.6%; twenty z=13 6.3% vs 0.4%; zone z=13 3.7% vs 0.0%; york z=13 5.2% vs 0.3%; london z=13 3.5% vs 0.1%; week z=12 6.8% vs 0.6%; month z=12 4.1% vs 0.2%; fall z=12 2.8% vs 0.0%; year z=11 3.9% vs 0.2%; the z=11 54.2% vs 30.1%; today z=10 8.1% vs 1.5%

Words, delete_account: account z=68 91.5% vs 3.5%; standard z=30 18.0% vs 0.6%; close z=28 14.4% vs 0.4%; delete z=23 11.9% vs 0.5%; platinum z=23 11.2% vs 0.4%; premium z=23 10.5% vs 0.4%; remove z=22 11.9% vs 0.6%; pro z=22 10.6% vs 0.4%; information z=22 13.7% vs 0.9%; cancel z=22 13.2% vs 0.9%; gold z=22 10.0% vs 0.4%; about z=18 19.0% vs 3.0%; freemium z=18 7.6% vs 0.3%; know z=15 19.8% vs 4.4%; cancellation z=15 5.5% vs 0.3%; problem z=15 7.1% vs 0.6%; deleting z=14 3.8% vs 0.0%; don't z=13 9.3% vs 1.5%; starter z=13 3.1% vs 0.1%; student z=13 3.1% vs 0.1%

Words, delivery_options: methods z=47 47.2% vs 0.5%; options z=44 42.4% vs 0.5%; shipment z=35 29.1% vs 0.2%; delivery z=34 34.6% vs 1.4%; shipping z=28 26.3% vs 1.4%; see z=26 37.5% vs 4.0%; what z=25 60.3% vs 11.3%; check z=20 28.5% vs 3.9%; choose z=19 8.4% vs 0.0%; are z=17 19.6% vs 2.4%; available z=17 10.2% vs 0.6%; offer z=16 10.8% vs 0.0%; help z=13 32.6% vs 9.5%; checking z=13 7.7% vs 0.7%; have z=12 23.4% vs 6.1%; there z=12 11.0% vs 1.7%; i z=11 68.2% vs 31.9%; seeing z=9 6.1% vs 0.8%; place z=9 3.7% vs 0.3%; calling z=9 2.2% vs 0.1%

Words, delivery_period: when z=40 31.9% vs 0.9%; how z=39 65.6% vs 7.1%; expect z=37 36.5% vs 0.1%; long z=35 22.2% vs 0.4%; see z=32 39.7% vs 3.8%; arrive z=30 48.3% vs 0.0%; soon z=27 36.6% vs 0.0%; check z=27 32.5% vs 3.7%; my z=26 64.0% vs 13.4%; can z=23 49.1% vs 9.8%; shipment z=22 10.3% vs 0.3%; item z=21 11.7% vs 0.7%; product z=19 11.3% vs 0.8%; article z=18 9.1% vs 0.5%; going z=18 10.3% vs 0.7%; will z=16 16.8% vs 2.5%; gonna z=16 5.8% vs 0.0%; takes z=15 21.2% vs 0.0%; checking z=15 7.6% vs 0.6%; i z=14 77.6% vs 31.5%

Words, edit_account: account z=53 68.1% vs 4.0%; on z=50 71.6% vs 5.9%; details z=34 20.2% vs 0.2%; edit z=31 16.5% vs 0.2%; user z=29 18.3% vs 0.8%; information z=29 18.8% vs 0.9%; info z=27 21.0% vs 0.0%; modify z=27 12.6% vs 0.3%; update z=23 11.3% vs 0.4%; editing z=22 8.6% vs 0.1%; data z=21 18.3% vs 0.0%; updating z=21 7.9% vs 0.1%; correct z=20 9.6% vs 0.5%; standard z=20 10.6% vs 0.7%; premium z=19 8.6% vs 0.4%; freemium z=19 8.1% vs 0.4%; correcting z=19 6.4% vs 0.1%; modifying z=19 6.4% vs 0.1%; platinum z=18 8.2% vs 0.5%; personal z=18 6.1% vs 0.2%

Words, email_addcontact: contacts z=21 43.3% vs 0.1%; email z=21 71.6% vs 1.2%; dot z=17 28.4% vs 0.1%; com z=17 28.4% vs 0.1%; contact z=16 40.3% vs 0.6%; add z=14 65.7% vs 3.5%; gmail z=13 16.4% vs 0.1%; save z=10 11.9% vs 0.1%; yahoo z=9 7.5% vs 0.0%; new z=8 31.3% vs 2.2%; sam z=6 4.5% vs 0.0%; my z=5 52.2% vs 14.1%; as z=5 6.0% vs 0.2%; john z=4 6.0% vs 0.3%; eagle z=4 1.5% vs 0.0%; dale z=4 1.5% vs 0.0%; doe z=4 1.5% vs 0.0%; cleveland z=4 1.5% vs 0.0%; sara z=4 1.5% vs 0.0%; jo z=3 1.5% vs 0.0%

Words, email_query: email z=41 33.7% vs 0.9%; any z=39 31.8% vs 1.0%; new z=34 31.9% vs 1.9%; emails z=25 47.2% vs 0.0%; from z=23 30.9% vs 4.9%; did z=17 8.9% vs 0.6%; received z=17 6.0% vs 0.0%; sent z=15 5.1% vs 0.0%; last z=15 9.5% vs 1.1%; recent z=15 4.2% vs 0.1%; check z=14 17.5% vs 4.0%; mail z=14 5.1% vs 0.3%; read z=14 4.6% vs 0.2%; subject z=13 3.7% vs 0.0%; messages z=12 3.7% vs 0.0%; if z=12 6.0% vs 0.6%; mom z=11 2.8% vs 0.1%; john z=11 3.7% vs 0.2%; gotten z=11 2.5% vs 0.1%; have z=11 18.2% vs 6.1%

Words, email_querycontact: number z=24 26.3% vs 0.1%; phone z=19 15.4% vs 0.1%; contact z=18 23.1% vs 0.6%; contacts z=17 12.8% vs 0.1%; john z=14 11.5% vs 0.2%; john's z=11 5.1% vs 0.0%; smith z=10 5.1% vs 0.0%; alex z=10 4.5% vs 0.0%; mary z=10 4.5% vs 0.0%; friend z=8 3.8% vs 0.1%; information z=8 10.3% vs 1.1%; home z=6 3.8% vs 0.2%; joe's z=6 2.6% vs 0.0%; is z=6 26.3% vs 9.5%; tara z=5 1.3% vs 0.0%; address z=5 12.8% vs 3.2%; mom's z=5 1.3% vs 0.0%; tell z=5 9.6% vs 1.9%; mom z=5 2.6% vs 0.1%; rahul z=5 1.3% vs 0.0%

Words, email_sendemail: email z=60 77.6% vs 0.5%; send z=49 49.1% vs 0.6%; mail z=20 9.1% vs 0.2%; an z=19 24.0% vs 3.2%; reply z=19 17.8% vs 0.0%; dot z=18 7.1% vs 0.1%; com z=18 6.9% vs 0.1%; gmail z=15 5.1% vs 0.0%; mom z=15 4.8% vs 0.1%; her z=15 4.6% vs 0.1%; to z=14 72.5% vs 31.6%; message z=14 4.2% vs 0.1%; john z=13 5.1% vs 0.2%; late z=12 3.0% vs 0.0%; friend z=12 3.0% vs 0.1%; write z=12 4.4% vs 0.2%; saying z=11 2.8% vs 0.0%; family z=11 3.6% vs 0.2%; asking z=11 2.6% vs 0.0%; following z=10 2.6% vs 0.1%

Words, general_affirm: command z=34 21.5% vs 0.3%; response z=30 17.1% vs 0.2%; you've z=27 14.5% vs 0.3%; right z=24 14.7% vs 0.6%; yes z=21 21.3% vs 0.0%; got z=21 14.7% vs 1.2%; understood z=19 9.6% vs 0.0%; answer z=18 7.7% vs 0.3%; instructions z=15 11.2% vs 0.0%; done z=15 6.5% vs 0.3%; gotten z=15 4.4% vs 0.0%; fantastic z=15 4.4% vs 0.1%; amazing z=14 4.0% vs 0.0%; that's z=14 6.1% vs 0.4%; you z=13 22.9% vs 8.2%; it's z=13 5.4% vs 0.3%; exact z=13 4.0% vs 0.0%; wonderful z=13 3.3% vs 0.1%; was z=13 7.5% vs 0.9%; perfectly z=12 14.7% vs 0.0%

Words, general_commandstop: olly z=39 49.8% vs 0.8%; just z=28 25.9% vs 0.4%; stop z=28 25.5% vs 0.1%; cancel z=27 29.9% vs 0.9%; you z=27 66.1% vs 8.0%; shut z=25 35.9% vs 0.0%; please z=23 45.8% vs 5.1%; down z=20 12.7% vs 0.2%; enough z=16 11.2% vs 0.0%; would z=16 17.9% vs 1.4%; that's z=14 9.6% vs 0.4%; off z=13 10.8% vs 0.7%; later z=13 4.8% vs 0.0%; all z=12 10.8% vs 0.9%; it z=11 22.7% vs 5.6%; will z=11 15.1% vs 2.7%; that z=9 12.4% vs 2.3%; up z=9 12.4% vs 2.6%; could z=8 10.4% vs 2.0%; talk z=7 4.8% vs 0.7%

Words, general_confirm: please z=48 77.9% vs 4.7%; check z=45 65.5% vs 3.6%; you z=40 77.9% vs 7.7%; said z=37 30.1% vs 0.5%; last z=36 32.0% vs 0.9%; confirm z=25 60.0% vs 0.0%; and z=25 30.8% vs 3.1%; thing z=22 12.8% vs 0.1%; details z=21 11.4% vs 0.3%; question z=21 10.0% vs 0.2%; detail z=19 13.5% vs 0.0%; would z=19 16.0% vs 1.4%; instruction z=17 9.4% vs 0.0%; like z=16 16.4% vs 2.1%; command z=15 7.8% vs 0.4%; correctly z=15 5.3% vs 0.1%; could z=13 13.0% vs 2.0%; just z=11 5.7% vs 0.5%; will z=11 12.6% vs 2.7%; i'd z=10 6.4% vs 0.8%

Words, general_dontcare: olly z=38 38.0% vs 0.8%; really z=26 32.2% vs 0.0%; mind z=24 27.8% vs 0.0%; not z=24 32.2% vs 3.1%; any z=22 20.6% vs 1.3%; no z=21 12.8% vs 0.4%; one z=20 20.6% vs 1.6%; would z=18 17.1% vs 1.4%; choose z=15 6.1% vs 0.1%; select z=15 6.7% vs 0.0%; matter z=15 18.0% vs 0.0%; care z=13 5.2% vs 0.0%; anything z=13 6.7% vs 0.3%; don't z=13 12.2% vs 1.5%; ok z=12 3.8% vs 0.0%; whatever z=12 9.9% vs 0.0%; does z=12 7.5% vs 0.6%; fine z=12 3.5% vs 0.0%; me z=12 35.4% vs 13.5%; doesn't z=11 10.4% vs 0.0%

Words, general_explain: please z=33 47.7% vs 4.8%; said z=32 22.6% vs 0.6%; again z=31 21.8% vs 0.6%; you z=27 49.0% vs 7.9%; clear z=27 14.5% vs 0.1%; your z=26 30.5% vs 3.1%; explain z=25 16.7% vs 0.1%; once z=22 10.3% vs 0.2%; not z=21 24.6% vs 3.1%; about z=20 22.6% vs 3.0%; just z=20 10.3% vs 0.4%; me z=20 49.4% vs 13.3%; on z=19 31.7% vs 6.4%; answers z=18 10.3% vs 0.0%; more z=18 10.5% vs 0.7%; so z=18 7.5% vs 0.2%; sentence z=18 7.3% vs 0.0%; words z=17 7.5% vs 0.0%; was z=17 10.6% vs 0.9%; answer z=16 7.0% vs 0.3%

Words, general_joke: joke z=15 70.7% vs 0.0%; funny z=14 17.4% vs 0.0%; tell z=12 31.5% vs 1.9%; jokes z=12 17.4% vs 0.0%; good z=8 10.9% vs 0.5%; laugh z=8 5.4% vs 0.0%; me z=6 43.5% vs 13.6%; a z=6 48.9% vs 18.4%; say z=5 5.4% vs 0.3%; heard z=5 2.2% vs 0.0%; lately z=5 2.2% vs 0.0%; hey z=5 4.3% vs 0.2%; hear z=4 6.5% vs 0.6%; banana z=4 1.1% vs 0.0%; i'll z=3 1.1% vs 0.0%; telling z=3 1.1% vs 0.0%; birds z=3 1.1% vs 0.0%; dirty z=3 1.1% vs 0.0%; make z=3 6.5% vs 1.2%; happen z=3 1.1% vs 0.0%

Words, general_negate: not z=45 45.1% vs 2.7%; think z=36 33.2% vs 0.1%; correct z=33 18.1% vs 0.4%; command z=31 15.6% vs 0.3%; right z=31 16.4% vs 0.5%; no z=30 14.4% vs 0.2%; wrong z=29 14.5% vs 0.1%; but z=29 27.4% vs 0.0%; you z=25 37.5% vs 7.9%; thanks z=24 9.6% vs 0.2%; answer z=24 9.8% vs 0.2%; response z=24 9.4% vs 0.2%; was z=23 12.5% vs 0.8%; did z=23 10.9% vs 0.6%; sorry z=21 18.8% vs 0.0%; that's z=19 7.3% vs 0.3%; didn't z=18 6.2% vs 0.0%; am z=17 12.8% vs 1.9%; that z=17 13.0% vs 2.2%; done z=16 5.8% vs 0.3%

Words, general_praise: olly z=46 33.2% vs 0.6%; much z=36 20.5% vs 0.2%; you z=36 50.9% vs 7.8%; done z=34 18.9% vs 0.2%; that's z=30 14.8% vs 0.2%; it's z=30 14.0% vs 0.2%; thanks z=29 13.3% vs 0.2%; so z=27 11.7% vs 0.2%; very z=24 10.0% vs 0.1%; work z=24 10.0% vs 0.3%; job z=23 13.0% vs 0.0%; that z=22 16.7% vs 2.1%; thank z=22 29.2% vs 0.0%; well z=21 10.4% vs 0.0%; helpful z=21 7.9% vs 0.1%; you've z=21 8.2% vs 0.3%; was z=20 10.6% vs 0.8%; nice z=20 6.4% vs 0.1%; correctly z=20 6.3% vs 0.1%; it z=20 23.6% vs 5.5%

Words, general_quirky: today z=20 11.5% vs 1.4%; day z=18 7.0% vs 0.5%; had z=13 2.7% vs 0.0%; tell z=12 8.1% vs 1.8%; would z=11 6.7% vs 1.4%; why z=11 1.9% vs 0.1%; hey z=11 2.5% vs 0.2%; you z=10 17.4% vs 8.2%; was z=10 4.7% vs 0.9%; happened z=10 1.5% vs 0.0%; if z=9 3.7% vs 0.7%; about z=9 8.6% vs 3.1%; life z=9 1.6% vs 0.1%; we z=9 1.8% vs 0.2%; explain z=8 1.9% vs 0.2%; are z=8 6.7% vs 2.5%; best z=8 2.5% vs 0.4%; able z=8 0.9% vs 0.0%; busy z=8 0.9% vs 0.0%; doing z=8 1.1% vs 0.1%

Words, general_repeat: again z=52 61.0% vs 0.3%; more z=43 38.2% vs 0.4%; please z=37 57.2% vs 4.8%; said z=35 27.0% vs 0.6%; one z=33 32.5% vs 1.5%; once z=30 19.5% vs 0.1%; say z=26 14.0% vs 0.1%; you z=25 47.8% vs 8.0%; hear z=24 15.1% vs 0.5%; try z=23 15.1% vs 0.7%; tell z=20 18.6% vs 1.8%; time z=20 19.5% vs 2.0%; last z=19 14.3% vs 1.0%; just z=18 10.1% vs 0.5%; would z=17 14.0% vs 1.4%; speak z=16 9.4% vs 0.6%; that z=14 14.5% vs 2.2%; let's z=14 4.4% vs 0.1%; i've z=13 3.5% vs 0.0%; us z=13 4.8% vs 0.2%

Words, get_invoice: invoice z=47 40.4% vs 0.5%; bill z=45 37.6% vs 0.7%; get z=38 36.3% vs 1.9%; from z=31 40.5% vs 4.7%; 12588 z=29 15.0% vs 0.2%; 85632 z=28 14.0% vs 0.2%; 37777 z=28 14.0% vs 0.2%; 00108 z=28 13.8% vs 0.2%; download z=24 35.1% vs 0.0%; know z=23 28.6% vs 4.2%; invoices z=22 9.0% vs 0.1%; bills z=22 8.9% vs 0.2%; do z=21 35.0% vs 7.2%; getting z=20 8.0% vs 0.1%; mail z=13 4.5% vs 0.2%; i z=13 67.9% vs 31.8%; don't z=12 9.2% vs 1.5%; my z=12 35.9% vs 13.9%; dont z=12 6.4% vs 0.8%; not z=11 13.1% vs 3.2%

Words, get_refund: money z=54 59.6% vs 0.3%; rebate z=30 16.0% vs 0.2%; of z=27 54.8% vs 11.8%; receive z=27 12.5% vs 0.2%; compensation z=25 11.1% vs 0.2%; refund z=23 12.2% vs 0.5%; reimbursement z=23 10.8% vs 0.4%; ask z=21 11.2% vs 0.6%; request z=21 10.0% vs 0.5%; reimbursements z=19 6.6% vs 0.2%; get z=19 16.3% vs 2.2%; refunds z=18 7.2% vs 0.3%; a z=17 51.4% vs 17.9%; restitution z=16 5.3% vs 0.2%; i z=14 66.4% vs 31.8%; receiving z=13 2.9% vs 0.1%; back z=13 4.9% vs 0.4%; my z=12 34.6% vs 13.9%; help z=12 26.0% vs 9.5%; paid z=11 10.4% vs 0.0%

Words, get_weather: weather z=61 42.9% vs 0.7%; in z=59 65.5% vs 8.0%; be z=58 37.8% vs 1.2%; will z=53 33.7% vs 1.6%; forecast z=37 27.6% vs 0.1%; it z=35 30.5% vs 4.8%; what's z=31 12.5% vs 0.7%; is z=31 36.4% vs 8.6%; now z=25 10.3% vs 1.0%; here z=24 7.0% vs 0.3%; park z=23 6.3% vs 0.1%; going z=21 7.0% vs 0.7%; state z=19 4.3% vs 0.1%; national z=18 3.9% vs 0.1%; there z=17 8.2% vs 1.5%; what z=16 27.9% vs 11.2%; like z=16 9.0% vs 1.9%; for z=15 26.5% vs 10.9%; the z=15 55.4% vs 29.4%; warm z=15 2.7% vs 0.0%

Words, iot_cleaning: start z=20 29.7% vs 0.4%; robot z=18 27.1% vs 0.0%; cleaning z=13 10.2% vs 0.0%; activate z=13 10.2% vs 0.0%; clean z=12 7.6% vs 0.0%; turn z=12 17.8% vs 0.8%; cleaner z=10 40.7% vs 0.0%; run z=8 4.2% vs 0.0%; initiate z=8 3.4% vs 0.0%; floor z=7 2.5% vs 0.0%; room z=7 5.9% vs 0.3%; on z=7 26.3% vs 6.6%; living z=6 4.2% vs 0.1%; dust z=6 2.5% vs 0.0%; vacuum z=6 72.9% vs 0.0%; the z=5 55.9% vs 30.3%; begin z=5 1.7% vs 0.0%; now z=4 7.6% vs 1.3%; house z=4 5.1% vs 0.6%; kitchen z=4 1.7% vs 0.1%

Words, iot_coffee: coffee z=28 88.9% vs 0.0%; make z=20 34.7% vs 1.1%; machine z=15 12.5% vs 0.0%; cup z=15 13.9% vs 0.0%; maker z=10 6.9% vs 0.0%; start z=10 10.4% vs 0.5%; pot z=9 4.2% vs 0.0%; please z=8 25.7% vs 5.2%; prepare z=8 3.5% vs 0.0%; brew z=7 5.6% vs 0.0%; some z=7 13.9% vs 2.2%; drink z=7 2.1% vs 0.0%; filter z=6 2.1% vs 0.0%; now z=5 8.3% vs 1.3%; starbucks z=5 1.4% vs 0.0%; me z=5 28.5% vs 13.6%; begin z=5 1.4% vs 0.0%; a z=4 34.0% vs 18.4%; set z=4 6.9% vs 1.5%; minutes z=4 2.8% vs 0.2%

Words, iot_hue_lightchange: lights z=34 50.6% vs 0.5%; change z=28 45.3% vs 0.9%; light z=27 32.4% vs 0.2%; blue z=21 21.2% vs 0.1%; red z=20 16.5% vs 0.1%; color z=20 25.3% vs 0.0%; green z=15 10.0% vs 0.1%; house z=13 14.1% vs 0.6%; room z=13 10.6% vs 0.2%; colour z=12 9.4% vs 0.0%; make z=12 16.5% vs 1.2%; yellow z=10 4.7% vs 0.0%; lighting z=10 4.1% vs 0.0%; living z=9 5.3% vs 0.1%; pink z=9 3.5% vs 0.0%; colors z=8 4.7% vs 0.0%; orange z=8 2.9% vs 0.0%; dark z=8 2.9% vs 0.0%; percent z=8 2.9% vs 0.0%; the z=7 60.0% vs 30.3%

Words, iot_hue_lightdim: lights z=31 69.4% vs 0.5%; lower z=21 29.6% vs 0.0%; down z=16 20.4% vs 0.2%; room z=13 15.3% vs 0.3%; brightness z=12 9.2% vs 0.0%; light z=11 14.3% vs 0.3%; living z=10 9.2% vs 0.1%; bright z=10 6.1% vs 0.0%; dim z=9 24.5% vs 0.0%; turn z=9 14.3% vs 0.8%; bit z=8 4.1% vs 0.0%; bedroom z=8 5.1% vs 0.1%; level z=7 4.1% vs 0.0%; reduce z=7 7.1% vs 0.0%; less z=7 3.1% vs 0.0%; intensity z=7 3.1% vs 0.0%; lighting z=7 3.1% vs 0.0%; the z=6 69.4% vs 30.3%; too z=6 4.1% vs 0.1%; percent z=6 3.1% vs 0.1%

Words, iot_hue_lightoff: off z=48 90.2% vs 0.4%; turn z=39 65.5% vs 0.6%; lights z=37 55.2% vs 0.5%; light z=29 33.3% vs 0.2%; room z=22 20.7% vs 0.2%; bedroom z=17 12.6% vs 0.0%; kitchen z=16 10.3% vs 0.0%; living z=14 9.2% vs 0.1%; bathroom z=12 6.9% vs 0.0%; lamp z=11 5.7% vs 0.0%; please z=9 24.1% vs 5.2%; bed z=8 2.9% vs 0.0%; switch z=8 6.3% vs 0.4%; side z=8 2.9% vs 0.0%; shut z=8 4.6% vs 0.2%; upstairs z=6 1.7% vs 0.0%; the z=6 53.4% vs 30.3%; garage z=6 1.7% vs 0.0%; in z=6 24.1% vs 10.0%; left z=5 1.7% vs 0.1%

Words, iot_hue_lighton: lights z=11 55.6% vs 0.6%; dark z=10 22.2% vs 0.0%; turn z=10 51.9% vs 0.8%; it's z=6 22.2% vs 0.4%; too z=6 11.1% vs 0.1%; on z=5 66.7% vs 6.6%; upstairs z=4 3.7% vs 0.0%; here z=4 14.8% vs 0.5%; corner z=4 3.7% vs 0.0%; hands z=3 3.7% vs 0.0%; light z=3 11.1% vs 0.4%; full z=3 3.7% vs 0.0%; brighter z=3 3.7% vs 0.0%; hall z=3 3.7% vs 0.0%; bright z=3 3.7% vs 0.0%; lamp z=3 3.7% vs 0.0%; disco z=3 3.7% vs 0.0%; can't z=2 7.4% vs 0.4%; all z=2 11.1% vs 0.9%; blue z=2 3.7% vs 0.1%

Words, iot_hue_lightup: lights z=26 50.5% vs 0.5%; light z=20 29.1% vs 0.3%; increase z=16 15.5% vs 0.0%; room z=16 19.4% vs 0.2%; brightness z=12 9.7% vs 0.0%; raise z=12 8.7% vs 0.0%; brighter z=11 11.7% vs 0.0%; up z=9 25.2% vs 2.6%; turn z=8 13.6% vs 0.8%; more z=8 12.6% vs 0.8%; living z=8 5.8% vs 0.1%; make z=7 12.6% vs 1.2%; lighting z=6 2.9% vs 0.0%; bright z=6 2.9% vs 0.0%; little z=6 3.9% vs 0.1%; please z=6 23.3% vs 5.3%; the z=6 62.1% vs 30.3%; turned z=6 1.9% vs 0.0%; intensity z=6 1.9% vs 0.0%; bit z=5 1.9% vs 0.0%

Words, iot_wemo_off: off z=29 84.0% vs 0.6%; socket z=25 60.0% vs 0.1%; plug z=24 48.0% vs 0.1%; turn z=21 54.7% vs 0.8%; wemo z=21 37.3% vs 0.0%; smart z=16 20.0% vs 0.0%; switch z=7 10.7% vs 0.4%; disable z=7 4.0% vs 0.0%; power z=7 5.3% vs 0.1%; shut z=6 6.7% vs 0.2%; active z=6 2.7% vs 0.0%; wifi z=5 2.7% vs 0.0%; down z=4 5.3% vs 0.3%; bedroom z=4 2.7% vs 0.1%; master z=4 1.3% vs 0.0%; turned z=4 1.3% vs 0.0%; mobile z=3 1.3% vs 0.0%; phone z=3 2.7% vs 0.1%; the z=3 53.3% vs 30.3%; rice z=3 1.3% vs 0.0%

Words, iot_wemo_on: plug z=23 50.0% vs 0.1%; socket z=22 45.2% vs 0.1%; wemo z=20 38.7% vs 0.1%; turn z=19 56.5% vs 0.8%; on z=13 82.3% vs 6.6%; smart z=11 12.9% vs 0.1%; power z=5 4.8% vs 0.1%; else z=5 3.2% vs 0.0%; switch z=4 6.5% vs 0.4%; turned z=4 1.6% vs 0.0%; button z=4 1.6% vs 0.0%; door z=4 1.6% vs 0.0%; start z=4 6.5% vs 0.5%; room z=4 4.8% vs 0.3%; mobile z=3 1.6% vs 0.0%; oven z=3 1.6% vs 0.0%; yourself z=3 1.6% vs 0.0%; computer z=3 1.6% vs 0.0%; rice z=3 1.6% vs 0.0%; then z=3 1.6% vs 0.0%

Words, lists_createoradd: list z=46 86.6% vs 1.6%; grocery z=21 15.7% vs 0.1%; new z=19 31.0% vs 2.1%; add z=19 37.5% vs 3.5%; create z=18 17.6% vs 0.6%; shopping z=14 8.8% vs 0.2%; milk z=13 6.0% vs 0.0%; make z=10 11.6% vs 1.2%; eggs z=10 3.2% vs 0.0%; start z=10 7.4% vs 0.5%; groceries z=8 2.3% vs 0.0%; put z=8 6.5% vs 0.6%; item z=7 6.5% vs 0.8%; a z=7 37.5% vs 18.3%; bananas z=6 1.4% vs 0.0%; shoes z=6 1.4% vs 0.0%; wash z=5 0.9% vs 0.0%; names z=5 0.9% vs 0.0%; created z=5 0.9% vs 0.0%; sugar z=5 0.9% vs 0.0%

Words, lists_query: list z=42 65.3% vs 1.6%; lists z=18 28.0% vs 0.0%; read z=15 8.5% vs 0.2%; shopping z=15 7.7% vs 0.2%; on z=14 32.1% vs 6.5%; my z=13 46.1% vs 14.0%; grocery z=12 5.2% vs 0.1%; items z=10 5.5% vs 0.3%; what z=9 32.5% vs 11.7%; are z=9 12.5% vs 2.6%; tell z=9 10.7% vs 1.9%; all z=8 6.6% vs 0.9%; available z=7 5.5% vs 0.7%; have z=7 18.5% vs 6.2%; names z=7 2.2% vs 0.0%; today z=7 7.7% vs 1.5%; what's z=7 6.3% vs 1.1%; whats z=7 3.0% vs 0.2%; do z=6 18.5% vs 7.5%; tasks z=6 1.1% vs 0.0%

Words, lists_remove: list z=45 81.2% vs 1.6%; remove z=36 46.0% vs 0.5%; delete z=26 27.2% vs 0.5%; item z=18 18.3% vs 0.7%; grocery z=17 9.8% vs 0.1%; from z=16 36.2% vs 5.0%; shopping z=15 9.4% vs 0.2%; erase z=11 4.0% vs 0.0%; rid z=10 3.1% vs 0.0%; apples z=8 2.7% vs 0.0%; bananas z=8 2.2% vs 0.0%; milk z=8 2.2% vs 0.0%; trash z=7 1.8% vs 0.0%; please z=7 17.4% vs 5.2%; eggs z=7 1.8% vs 0.0%; my z=6 29.0% vs 14.1%; notes z=6 1.3% vs 0.0%; groceries z=6 1.3% vs 0.0%; bread z=6 1.3% vs 0.0%; off z=6 4.9% vs 0.7%

Words, music_likeness: song z=26 54.5% vs 1.4%; save z=22 27.3% vs 0.1%; this z=14 40.9% vs 4.1%; favorite z=12 9.8% vs 0.1%; love z=11 10.6% vs 0.3%; favorites z=11 6.1% vs 0.0%; remember z=9 5.3% vs 0.1%; music z=9 17.4% vs 1.7%; like z=9 18.9% vs 2.1%; enjoy z=8 3.0% vs 0.0%; opinion z=7 5.3% vs 0.2%; songs z=7 6.8% vs 0.4%; that z=6 13.6% vs 2.3%; preferences z=6 3.0% vs 0.0%; favourite z=6 2.3% vs 0.0%; rock z=6 5.3% vs 0.3%; really z=6 4.5% vs 0.2%; singer z=6 2.3% vs 0.0%; note z=6 2.3% vs 0.0%; voice z=5 1.5% vs 0.0%

Words, music_query: song z=33 54.2% vs 1.3%; name z=18 13.4% vs 0.1%; who z=17 13.9% vs 0.3%; playing z=16 19.4% vs 1.1%; this z=15 33.3% vs 4.1%; is z=13 45.3% vs 9.4%; music z=13 18.4% vs 1.6%; artist z=12 9.0% vs 0.4%; currently z=11 5.0% vs 0.1%; singer z=11 4.5% vs 0.0%; title z=11 4.5% vs 0.0%; genre z=9 3.5% vs 0.0%; what z=9 37.3% vs 11.7%; listening z=9 3.0% vs 0.0%; band z=9 3.0% vs 0.0%; singing z=8 3.0% vs 0.0%; sings z=7 2.0% vs 0.0%; most z=7 4.0% vs 0.2%; wrote z=7 2.0% vs 0.0%; tell z=7 10.4% vs 1.9%

Words, music_settings: repeat z=18 33.9% vs 0.1%; shuffle z=16 24.2% vs 0.0%; song z=11 35.5% vs 1.5%; settings z=7 4.8% vs 0.0%; mode z=7 4.8% vs 0.0%; music z=6 21.0% vs 1.7%; skip z=6 4.8% vs 0.0%; player z=6 4.8% vs 0.1%; please z=5 27.4% vs 5.3%; loop z=4 1.6% vs 0.0%; songs z=4 6.5% vs 0.5%; required z=4 1.6% vs 0.0%; lie z=4 1.6% vs 0.0%; playlists z=4 1.6% vs 0.0%; volume z=4 4.8% vs 0.3%; lower z=4 3.2% vs 0.1%; bass z=4 1.6% vs 0.0%; gone z=3 1.6% vs 0.0%; setting z=3 4.8% vs 0.3%; this z=3 17.7% vs 4.2%

Words, news_query: news z=48 53.9% vs 0.2%; latest z=35 20.2% vs 0.2%; world z=20 6.7% vs 0.2%; trump z=18 5.9% vs 0.0%; on z=16 22.5% vs 6.4%; what's z=16 8.8% vs 1.0%; updates z=16 4.4% vs 0.1%; happening z=15 4.2% vs 0.1%; bbc z=14 3.4% vs 0.0%; me z=13 30.5% vs 13.4%; about z=13 12.6% vs 3.1%; the z=13 51.0% vs 30.1%; new z=12 9.7% vs 2.1%; tell z=12 8.9% vs 1.8%; today's z=12 2.8% vs 0.1%; read z=12 3.4% vs 0.2%; today z=11 7.3% vs 1.4%; stories z=11 2.0% vs 0.0%; election z=11 2.5% vs 0.0%; donald z=11 1.9% vs 0.0%

Words, newsletter_subscription: company z=38 25.1% vs 0.8%; your z=31 29.4% vs 2.9%; receive z=28 12.4% vs 0.2%; cancel z=27 15.5% vs 0.8%; sign z=27 13.8% vs 0.6%; ur z=24 15.8% vs 1.3%; to z=19 74.8% vs 31.3%; up z=19 16.1% vs 2.4%; help z=17 31.1% vs 9.4%; newsletter z=17 94.3% vs 0.0%; receiving z=15 3.7% vs 0.0%; canceling z=13 3.6% vs 0.2%; cancelling z=13 3.6% vs 0.2%; signing z=11 2.3% vs 0.0%; i z=11 56.7% vs 31.9%; about z=10 10.8% vs 3.1%; subscribe z=9 9.8% vs 0.0%; need z=9 17.1% vs 7.4%; the z=8 49.0% vs 30.1%; subscription z=8 28.3% vs 0.0%

Words, payment_issue: payment z=60 62.7% vs 1.5%; with z=58 76.9% vs 4.3%; inform z=36 22.8% vs 0.4%; report z=36 22.6% vs 0.4%; notify z=36 22.4% vs 0.4%; online z=29 35.7% vs 0.0%; of z=28 56.7% vs 11.8%; issues z=28 14.2% vs 0.4%; error z=27 13.1% vs 0.3%; problems z=27 14.3% vs 0.5%; issue z=26 13.1% vs 0.4%; errors z=26 12.2% vs 0.3%; troubles z=23 9.5% vs 0.1%; trouble z=23 9.1% vs 0.1%; problem z=22 11.1% vs 0.5%; reporting z=18 5.8% vs 0.1%; notifying z=18 5.7% vs 0.1%; informing z=17 5.0% vs 0.1%; an z=17 17.7% vs 3.2%; help z=15 30.7% vs 9.4%

Words, place_order: several z=38 29.1% vs 0.2%; some z=38 30.9% vs 1.8%; article z=38 22.7% vs 0.3%; product z=31 18.0% vs 0.7%; item z=31 17.0% vs 0.6%; your z=28 26.7% vs 3.0%; articles z=26 11.2% vs 0.1%; buy z=26 11.2% vs 0.1%; shop z=25 12.5% vs 0.1%; items z=23 9.0% vs 0.2%; products z=22 9.5% vs 0.4%; ur z=21 13.2% vs 1.3%; of z=20 41.0% vs 12.0%; purchase z=20 13.7% vs 1.6%; order z=18 15.0% vs 2.3%; few z=15 12.6% vs 0.0%; do z=15 24.3% vs 7.3%; know z=15 17.8% vs 4.4%; i z=15 64.2% vs 31.7%; help z=14 27.0% vs 9.5%

Words, play_audiobook: audio z=16 14.0% vs 0.0%; resume z=16 25.5% vs 0.0%; left z=13 7.0% vs 0.0%; play z=12 34.4% vs 5.4%; book z=11 25.5% vs 3.1%; continue z=11 7.0% vs 0.0%; reading z=11 5.1% vs 0.0%; harry z=10 4.5% vs 0.0%; potter z=9 4.5% vs 0.0%; start z=9 8.9% vs 0.5%; pause z=9 3.8% vs 0.0%; last z=8 10.8% vs 1.1%; war z=7 3.2% vs 0.1%; towers z=7 2.5% vs 0.0%; playback z=7 3.8% vs 0.0%; chapter z=7 1.9% vs 0.0%; gods z=7 1.9% vs 0.0%; farm z=7 1.9% vs 0.0%; off z=6 7.0% vs 0.7%; restart z=6 3.2% vs 0.0%

Words, play_game: game z=28 33.3% vs 0.4%; play z=26 67.2% vs 5.2%; let's z=16 9.8% vs 0.1%; games z=12 5.7% vs 0.0%; lets z=11 5.2% vs 0.0%; open z=10 10.9% vs 0.8%; bird z=10 4.0% vs 0.0%; start z=9 8.0% vs 0.5%; with z=9 23.6% vs 5.4%; angry z=9 2.9% vs 0.0%; poker z=8 2.3% vs 0.0%; temple z=7 2.3% vs 0.0%; together z=7 2.3% vs 0.0%; football z=7 2.3% vs 0.0%; never z=7 2.3% vs 0.0%; run z=7 2.3% vs 0.0%; candy z=7 3.4% vs 0.0%; crush z=7 3.4% vs 0.0%; subway z=7 1.7% vs 0.0%; mario z=7 1.7% vs 0.0%

Words, play_music: play z=98 81.1% vs 1.3%; by z=56 23.6% vs 0.8%; music z=53 26.0% vs 0.4%; from z=42 22.6% vs 4.2%; song z=36 11.3% vs 1.0%; some z=34 12.4% vs 1.7%; on z=32 20.5% vs 5.9%; hear z=30 7.0% vs 0.3%; songs z=28 6.5% vs 0.1%; top z=24 4.5% vs 0.3%; playlist z=21 8.5% vs 2.4%; listen z=20 3.2% vs 0.1%; track z=20 4.1% vs 0.5%; album z=18 4.1% vs 0.7%; google z=17 2.3% vs 0.1%; spotify z=16 2.2% vs 0.0%; rock z=16 2.3% vs 0.2%; popular z=16 1.9% vs 0.1%; pandora z=15 1.8% vs 0.0%; jazz z=13 1.4% vs 0.1%

Words, play_podcasts: next z=28 30.9% vs 0.9%; play z=22 44.6% vs 5.2%; episode z=19 32.7% vs 0.0%; podcast z=15 65.1% vs 0.0%; start z=13 8.9% vs 0.5%; skip z=12 5.6% vs 0.0%; forward z=11 3.7% vs 0.0%; go z=10 7.1% vs 0.5%; previous z=10 3.0% vs 0.0%; favourite z=9 2.2% vs 0.0%; move z=9 2.2% vs 0.0%; played z=8 2.2% vs 0.1%; listen z=7 3.7% vs 0.3%; please z=7 14.9% vs 5.2%; npr z=7 1.5% vs 0.0%; latest z=7 4.1% vs 0.5%; rewind z=6 1.1% vs 0.0%; guys z=6 1.1% vs 0.0%; beginning z=6 1.1% vs 0.0%; most z=6 2.6% vs 0.2%

Words, play_radio: radio z=34 55.7% vs 0.1%; play z=28 44.8% vs 5.1%; station z=25 15.1% vs 0.1%; fm z=23 12.1% vs 0.1%; point z=22 11.6% vs 0.1%; turn z=20 13.9% vs 0.7%; ninety z=19 9.8% vs 0.0%; on z=18 32.7% vs 6.5%; hundred z=17 7.8% vs 0.2%; nine z=17 8.8% vs 0.3%; channel z=16 9.3% vs 0.0%; listen z=14 6.3% vs 0.3%; seven z=14 7.1% vs 0.4%; pandora z=13 4.5% vs 0.1%; one z=12 11.6% vs 1.7%; start z=12 6.0% vs 0.5%; and z=12 15.1% vs 3.3%; bbc z=11 2.8% vs 0.1%; npr z=10 3.0% vs 0.0%; eighty z=10 2.3% vs 0.0%

Words, qa_currency: rate z=31 44.7% vs 1.7%; dollars z=28 25.4% vs 0.1%; us z=25 19.0% vs 0.2%; s z=23 18.0% vs 0.1%; between z=23 16.2% vs 0.1%; d z=22 18.0% vs 0.0%; u z=21 16.9% vs 0.0%; much z=18 13.7% vs 0.4%; dollar z=17 33.1% vs 0.0%; many z=17 12.0% vs 0.3%; exchange z=17 46.5% vs 0.0%; is z=15 45.1% vs 9.3%; worth z=12 4.6% vs 0.0%; rates z=12 7.4% vs 0.0%; and z=12 20.8% vs 3.3%; pound z=12 7.0% vs 0.0%; convert z=12 4.2% vs 0.0%; p z=11 3.5% vs 0.0%; g z=11 3.5% vs 0.0%; indian z=10 3.5% vs 0.0%

Words, qa_definition: does z=22 15.2% vs 0.6%; mean z=20 10.5% vs 0.0%; word z=20 13.1% vs 0.0%; what z=18 42.3% vs 11.6%; describe z=17 14.0% vs 0.0%; meaning z=14 5.5% vs 0.0%; define z=12 13.7% vs 0.0%; description z=11 4.7% vs 0.0%; is z=10 22.7% vs 9.4%; look z=10 6.1% vs 0.8%; definition z=10 17.8% vs 0.0%; of z=10 25.4% vs 12.4%; computer z=9 2.0% vs 0.0%; object z=9 2.3% vs 0.0%; tell z=9 8.2% vs 1.9%; looks z=7 1.5% vs 0.0%; laptop z=7 1.2% vs 0.0%; siri z=7 1.2% vs 0.0%; banana z=6 0.9% vs 0.0%; ultra z=6 0.9% vs 0.0%

Words, qa_factoid: is z=34 47.2% vs 9.0%; many z=23 8.3% vs 0.3%; who z=19 6.2% vs 0.2%; was z=17 7.9% vs 0.9%; old z=17 4.3% vs 0.1%; how z=16 22.9% vs 7.7%; where z=15 11.7% vs 2.5%; world z=15 4.0% vs 0.2%; born z=13 3.9% vs 0.0%; what z=13 24.8% vs 11.6%; tell z=12 8.1% vs 1.9%; located z=12 2.7% vs 0.0%; did z=12 4.7% vs 0.7%; president z=11 2.0% vs 0.0%; states z=11 1.9% vs 0.0%; when z=11 5.9% vs 1.3%; does z=11 4.0% vs 0.6%; highest z=10 1.6% vs 0.0%; mountain z=10 1.7% vs 0.1%; jackson z=10 1.6% vs 0.0%

Words, qa_maths: plus z=21 27.0% vs 0.1%; two z=14 24.3% vs 0.9%; answer z=11 13.5% vs 0.4%; hundred z=11 10.8% vs 0.2%; one z=10 23.4% vs 1.7%; five z=10 17.1% vs 0.9%; square z=10 6.3% vs 0.0%; is z=10 51.4% vs 9.4%; four z=9 13.5% vs 0.7%; times z=9 11.7% vs 0.5%; math z=9 5.4% vs 0.0%; divided z=9 8.1% vs 0.0%; what z=8 48.6% vs 11.7%; fifty z=8 5.4% vs 0.1%; twenty z=8 9.9% vs 0.4%; three z=8 11.7% vs 0.7%; calculate z=8 3.6% vs 0.0%; sixty z=7 3.6% vs 0.0%; thirteen z=7 3.6% vs 0.0%; fourteen z=7 3.6% vs 0.0%

Words, qa_stock: price z=28 38.1% vs 0.1%; stock z=18 71.1% vs 0.0%; apple z=14 7.6% vs 0.0%; market z=14 7.6% vs 0.0%; prices z=14 10.2% vs 0.0%; google z=11 7.1% vs 0.2%; share z=11 4.6% vs 0.0%; amazon z=10 4.1% vs 0.1%; microsoft z=9 3.6% vs 0.0%; walmart z=9 3.0% vs 0.0%; much z=8 6.6% vs 0.5%; is z=8 29.9% vs 9.5%; stocks z=8 8.1% vs 0.0%; cost z=8 2.5% vs 0.0%; closing z=8 2.5% vs 0.0%; doing z=8 3.0% vs 0.1%; worth z=7 2.5% vs 0.1%; value z=7 3.0% vs 0.1%; m z=7 3.0% vs 0.1%; b z=7 3.0% vs 0.1%

Words, rate_book: rate z=56 50.5% vs 0.3%; give z=55 33.7% vs 0.9%; out z=52 35.0% vs 0.4%; 6 z=50 54.0% vs 0.2%; of z=48 60.5% vs 10.8%; current z=42 21.9% vs 1.2%; this z=41 29.8% vs 3.3%; stars z=30 29.2% vs 0.0%; four z=29 10.1% vs 0.4%; 1 z=29 9.7% vs 0.3%; 3 z=27 8.5% vs 0.2%; 5 z=27 8.7% vs 0.2%; novel z=27 8.7% vs 0.1%; 4 z=27 8.1% vs 0.2%; 2 z=25 7.4% vs 0.3%; three z=24 7.9% vs 0.5%; book z=23 14.9% vs 2.8%; five z=22 7.8% vs 0.7%; series z=22 5.6% vs 0.2%; two z=21 7.6% vs 0.7%

Words, recommendation_events: events z=34 39.3% vs 0.4%; happening z=23 16.6% vs 0.1%; weekend z=20 13.0% vs 0.1%; area z=20 16.2% vs 0.4%; local z=18 11.7% vs 0.1%; in z=16 49.0% vs 9.8%; going z=14 14.2% vs 0.8%; tonight z=14 7.3% vs 0.1%; around z=13 8.9% vs 0.3%; any z=13 15.4% vs 1.3%; are z=13 20.2% vs 2.5%; event z=11 10.1% vs 0.7%; town z=11 4.5% vs 0.1%; there z=10 12.6% vs 1.7%; near z=9 4.9% vs 0.2%; this z=9 19.0% vs 4.1%; nearby z=8 4.5% vs 0.2%; recommend z=8 2.4% vs 0.0%; concerts z=8 3.2% vs 0.0%; interesting z=8 2.0% vs 0.0%

Words, recommendation_locations: near z=17 13.1% vs 0.2%; where z=14 26.7% vs 2.5%; restaurants z=14 8.4% vs 0.0%; local z=13 8.9% vs 0.2%; shops z=13 14.1% vs 0.0%; shop z=13 9.4% vs 0.2%; area z=12 9.9% vs 0.4%; recommend z=12 5.2% vs 0.0%; good z=11 9.9% vs 0.4%; nearby z=11 7.3% vs 0.2%; nearest z=10 7.9% vs 0.3%; around z=10 7.3% vs 0.3%; best z=10 8.4% vs 0.4%; clothes z=10 3.7% vs 0.0%; place z=9 6.8% vs 0.3%; restaurant z=9 12.6% vs 1.4%; italian z=9 3.1% vs 0.0%; places z=9 3.1% vs 0.0%; store z=9 3.7% vs 0.1%; shopping z=8 4.7% vs 0.2%

Words, recommendation_movies: movies z=17 39.0% vs 0.8%; movie z=12 36.6% vs 1.8%; playing z=11 25.6% vs 1.1%; are z=9 29.3% vs 2.6%; good z=8 13.4% vs 0.5%; recommend z=8 6.1% vs 0.0%; watch z=8 9.8% vs 0.3%; near z=7 8.5% vs 0.2%; what z=7 51.2% vs 11.7%; tonight z=6 6.1% vs 0.2%; action z=5 2.4% vs 0.0%; suggest z=5 2.4% vs 0.0%; reviews z=5 2.4% vs 0.0%; shown z=5 2.4% vs 0.0%; released z=5 2.4% vs 0.0%; theaters z=5 3.7% vs 0.1%; should z=5 6.1% vs 0.3%; amc z=4 2.4% vs 0.0%; highest z=4 2.4% vs 0.0%; theatre z=4 4.9% vs 0.3%

Words, recover_password: user z=48 42.2% vs 0.3%; my z=44 88.7% vs 13.0%; profile z=40 26.5% vs 0.4%; account z=26 31.3% vs 4.3%; reset z=26 21.5% vs 0.0%; key z=21 31.3% vs 0.0%; code z=20 16.3% vs 0.0%; access z=18 15.1% vs 0.0%; lost z=18 5.5% vs 0.1%; pass z=16 6.7% vs 0.0%; of z=15 37.0% vs 12.1%; information z=14 8.1% vs 1.0%; help z=12 26.6% vs 9.5%; i z=11 63.4% vs 31.8%; retrieve z=10 18.7% vs 0.0%; about z=10 11.1% vs 3.1%; question z=9 2.7% vs 0.2%; pwd z=9 9.3% vs 0.0%; pin z=8 35.0% vs 0.0%; remember z=8 1.4% vs 0.1%

Words, registration_problems: sign z=43 35.8% vs 0.2%; with z=42 53.8% vs 4.7%; up z=39 36.5% vs 2.1%; inform z=39 25.9% vs 0.4%; report z=38 24.4% vs 0.4%; notify z=38 24.0% vs 0.4%; issue z=34 19.4% vs 0.3%; of z=31 61.1% vs 11.7%; issues z=30 15.9% vs 0.3%; error z=30 15.1% vs 0.3%; errors z=29 14.4% vs 0.3%; problems z=28 15.4% vs 0.5%; problem z=28 14.5% vs 0.5%; an z=22 22.2% vs 3.1%; notifying z=18 5.3% vs 0.1%; cannot z=16 6.8% vs 0.4%; informing z=16 4.5% vs 0.1%; reporting z=16 4.5% vs 0.1%; a z=15 47.6% vs 17.9%; i z=12 63.3% vs 31.8%

Words, review: about z=38 38.7% vs 2.6%; leave z=34 33.3% vs 0.1%; send z=32 21.0% vs 0.8%; product z=32 20.0% vs 0.7%; submit z=31 16.6% vs 0.3%; products z=30 15.8% vs 0.3%; your z=29 29.6% vs 2.9%; service z=28 17.2% vs 0.8%; company z=28 17.8% vs 0.9%; ur z=28 19.8% vs 1.2%; write z=25 11.2% vs 0.1%; for z=24 48.1% vs 10.9%; review z=23 15.6% vs 0.0%; some z=22 18.2% vs 2.0%; services z=22 14.7% vs 0.0%; a z=20 58.8% vs 17.8%; opinion z=17 11.8% vs 0.0%; an z=16 17.7% vs 3.2%; feedback z=16 54.9% vs 0.0%; comment z=15 14.7% vs 0.0%

Words, search_creative_work: find z=69 45.7% vs 1.9%; show z=53 28.0% vs 1.3%; the z=35 68.6% vs 29.0%; called z=35 15.5% vs 0.1%; game z=30 9.1% vs 0.2%; saga z=23 5.0% vs 0.2%; tv z=22 9.9% vs 0.0%; search z=21 4.5% vs 0.1%; series z=21 4.4% vs 0.2%; soundtrack z=20 4.0% vs 0.1%; picture z=18 4.9% vs 0.0%; novel z=18 3.8% vs 0.3%; looking z=17 3.4% vs 0.2%; video z=17 4.2% vs 0.0%; look z=16 4.7% vs 0.7%; watch z=15 2.7% vs 0.2%; where z=14 7.6% vs 2.4%; of z=14 21.7% vs 12.1%; television z=14 4.6% vs 0.0%; album z=13 4.0% vs 0.8%

Words, search_screening_event: movie z=55 48.3% vs 0.3%; playing z=45 28.3% vs 0.2%; at z=44 37.9% vs 4.3%; movies z=37 24.0% vs 0.1%; schedule z=37 15.8% vs 0.3%; find z=31 21.5% vs 2.8%; times z=29 10.1% vs 0.2%; house z=28 9.3% vs 0.3%; are z=26 15.8% vs 2.2%; nearest z=25 8.9% vs 0.1%; is z=22 29.4% vs 8.9%; closest z=22 8.3% vs 0.0%; time z=21 11.7% vs 1.9%; the z=21 63.4% vs 29.3%; showing z=20 8.9% vs 0.0%; watch z=19 4.4% vs 0.1%; theatre z=19 7.9% vs 0.0%; what z=19 30.4% vs 11.2%; schedules z=18 10.5% vs 0.0%; theatres z=17 17.0% vs 0.0%

Words, set_up_shipping_address: address z=78 97.0% vs 1.7%; shipping z=55 49.4% vs 0.8%; delivery z=53 46.9% vs 1.0%; up z=49 47.1% vs 1.9%; set z=38 28.3% vs 1.1%; submit z=31 15.9% vs 0.3%; new z=31 25.3% vs 1.9%; another z=21 7.8% vs 0.2%; setting z=20 18.5% vs 0.0%; submitting z=20 8.2% vs 0.0%; enter z=20 14.9% vs 0.0%; different z=19 6.9% vs 0.0%; secondary z=13 24.4% vs 0.0%; there z=11 8.3% vs 1.6%; have z=11 18.3% vs 6.1%; i z=11 59.6% vs 31.8%; why z=10 2.0% vs 0.1%; cannot z=10 4.0% vs 0.5%; troubles z=10 2.6% vs 0.2%; trouble z=9 2.5% vs 0.2%

Words, social_post: complaint z=30 23.4% vs 0.5%; post z=25 16.9% vs 0.1%; service z=24 20.2% vs 0.9%; facebook z=24 13.6% vs 0.1%; twitter z=21 11.8% vs 0.1%; their z=16 6.3% vs 0.0%; tweet z=16 46.9% vs 0.0%; send z=14 10.8% vs 1.0%; consumer z=12 6.5% vs 0.4%; amazon z=11 3.0% vs 0.0%; bad z=11 3.0% vs 0.1%; write z=10 4.3% vs 0.2%; that z=10 11.8% vs 2.2%; they z=10 2.8% vs 0.1%; them z=10 3.0% vs 0.1%; about z=10 14.1% vs 3.1%; message z=10 2.8% vs 0.1%; samsung z=9 2.5% vs 0.0%; instagram z=9 1.8% vs 0.0%; status z=9 7.1% vs 1.2%

Words, social_query: facebook z=27 34.5% vs 0.1%; twitter z=17 15.2% vs 0.1%; media z=17 15.2% vs 0.0%; social z=16 15.9% vs 0.0%; on z=12 44.1% vs 6.6%; trending z=12 6.9% vs 0.0%; instagram z=10 5.5% vs 0.0%; feed z=10 6.2% vs 0.0%; post z=10 7.6% vs 0.2%; anything z=9 9.0% vs 0.4%; anyone z=9 4.1% vs 0.0%; updates z=9 6.2% vs 0.2%; likes z=9 4.8% vs 0.0%; what's z=9 13.1% vs 1.1%; posted z=9 4.1% vs 0.0%; latest z=9 9.0% vs 0.5%; friends z=9 4.1% vs 0.0%; new z=8 17.2% vs 2.2%; happening z=8 5.5% vs 0.2%; today z=7 12.4% vs 1.5%

Words, switch_account: account z=53 70.6% vs 3.9%; use z=40 29.9% vs 0.2%; change z=39 29.4% vs 0.7%; profile z=35 23.5% vs 0.6%; switch z=29 15.7% vs 0.2%; standard z=28 17.0% vs 0.6%; the z=24 86.9% vs 29.6%; pro z=23 11.4% vs 0.4%; gold z=23 11.2% vs 0.4%; premium z=21 10.2% vs 0.4%; platinum z=21 10.4% vs 0.4%; freemium z=21 9.6% vs 0.3%; using z=20 10.6% vs 0.0%; to z=19 77.9% vs 31.4%; switching z=16 4.6% vs 0.0%; changing z=15 5.0% vs 0.2%; help z=13 28.8% vs 9.5%; enterprise z=12 3.0% vs 0.1%; silver z=11 3.0% vs 0.1%; personal z=11 3.4% vs 0.2%

Words, takeaway_order: order z=30 72.6% vs 2.3%; pizza z=22 24.7% vs 0.1%; food z=16 17.1% vs 0.3%; chinese z=14 8.9% vs 0.0%; takeout z=13 11.6% vs 0.0%; takeaway z=11 6.8% vs 0.1%; from z=9 29.5% vs 5.1%; near z=9 6.8% vs 0.2%; cheese z=9 4.1% vs 0.0%; call z=8 9.6% vs 0.6%; pizzas z=8 3.4% vs 0.0%; china z=8 3.4% vs 0.0%; domino's z=7 2.7% vs 0.0%; wings z=7 2.7% vs 0.0%; dominos z=7 2.7% vs 0.0%; dominoes z=7 3.4% vs 0.0%; pepperoni z=7 3.4% vs 0.0%; ham z=7 2.7% vs 0.0%; place z=7 6.2% vs 0.4%; some z=7 13.7% vs 2.2%

Words, takeaway_query: does z=21 27.1% vs 0.6%; takeaway z=18 22.3% vs 0.0%; order z=14 27.1% vs 2.4%; deliver z=13 10.2% vs 0.0%; pizza z=13 8.4% vs 0.1%; food z=12 10.8% vs 0.3%; restaurant z=12 18.7% vs 1.4%; garden z=11 5.4% vs 0.0%; ready z=10 4.2% vs 0.0%; delivery z=9 15.1% vs 1.7%; hut z=9 3.6% vs 0.0%; takeout z=9 3.6% vs 0.0%; if z=8 9.0% vs 0.7%; olive z=8 3.0% vs 0.0%; take z=8 7.8% vs 0.5%; they z=8 3.6% vs 0.1%; orders z=8 3.0% vs 0.0%; delivered z=8 3.6% vs 0.0%; chinese z=8 3.0% vs 0.0%; has z=7 6.0% vs 0.4%

Words, track_order: status z=50 44.4% vs 0.7%; order z=48 50.2% vs 1.9%; purchase z=48 44.7% vs 1.3%; ord z=35 23.7% vs 0.5%; see z=30 35.6% vs 4.0%; of z=26 53.5% vs 11.9%; current z=22 17.9% vs 1.6%; check z=21 24.8% vs 3.9%; locate z=18 6.7% vs 0.2%; checking z=14 7.3% vs 0.7%; help z=13 28.1% vs 9.5%; seeing z=12 6.7% vs 0.8%; eta z=12 35.2% vs 0.0%; track z=11 5.5% vs 0.6%; locating z=9 1.5% vs 0.0%; the z=9 49.3% vs 30.1%; i z=8 50.9% vs 32.0%; wanna z=8 3.6% vs 0.5%; tracking z=8 1.1% vs 0.0%; need z=7 16.2% vs 7.4%

Words, track_refund: status z=46 45.0% vs 0.7%; if z=35 26.8% vs 0.4%; rebate z=33 22.6% vs 0.2%; compensation z=30 19.7% vs 0.2%; refund z=29 20.0% vs 0.5%; there z=29 27.4% vs 1.5%; restitution z=27 19.5% vs 0.1%; reimbursement z=26 16.0% vs 0.3%; anything z=23 11.8% vs 0.3%; see z=21 30.5% vs 4.1%; current z=21 19.7% vs 1.6%; any z=19 16.0% vs 1.2%; updates z=18 7.4% vs 0.1%; waiting z=17 8.3% vs 0.0%; dollars z=17 6.8% vs 0.2%; check z=16 23.3% vs 3.9%; expect z=15 8.6% vs 0.6%; are z=14 15.8% vs 2.5%; track z=14 7.9% vs 0.6%; news z=13 8.8% vs 0.8%

Words, transport_query: train z=45 57.8% vs 0.3%; does z=17 13.3% vs 0.6%; ticket z=15 9.0% vs 0.3%; station z=15 7.3% vs 0.2%; times z=14 10.3% vs 0.5%; time z=14 19.3% vs 2.1%; chicago z=13 5.3% vs 0.1%; route z=11 4.0% vs 0.0%; trains z=11 5.0% vs 0.0%; directions z=10 8.6% vs 0.0%; york z=10 5.3% vs 0.3%; when z=10 10.3% vs 1.3%; next z=9 9.0% vs 1.0%; get z=9 13.3% vs 2.3%; travel z=9 2.7% vs 0.0%; location z=9 4.0% vs 0.2%; leave z=9 6.3% vs 0.6%; tickets z=8 2.3% vs 0.0%; orlando z=8 2.0% vs 0.0%; from z=8 17.9% vs 5.1%

Words, transport_taxi: book z=14 37.2% vs 3.1%; airport z=14 11.7% vs 0.0%; uber z=12 31.4% vs 0.0%; pick z=12 7.3% vs 0.0%; call z=11 14.6% vs 0.6%; ride z=10 5.8% vs 0.0%; cab z=10 14.6% vs 0.0%; taxi z=10 53.3% vs 0.0%; an z=8 21.9% vs 3.4%; a z=8 57.7% vs 18.3%; home z=8 5.8% vs 0.2%; mall z=7 2.9% vs 0.0%; me z=7 45.3% vs 13.6%; golden z=6 2.2% vs 0.0%; now z=6 10.2% vs 1.3%; car z=6 2.9% vs 0.1%; america z=5 2.2% vs 0.0%; location z=5 3.6% vs 0.2%; hour z=5 3.6% vs 0.3%; go z=5 5.1% vs 0.6%

Words, transport_ticket: train z=45 89.1% vs 0.4%; ticket z=33 78.7% vs 0.1%; book z=21 49.4% vs 3.0%; york z=16 14.9% vs 0.3%; tickets z=12 6.9% vs 0.0%; boston z=11 5.7% vs 0.0%; mumbai z=11 5.2% vs 0.0%; a z=10 63.2% vs 18.3%; buy z=9 6.9% vs 0.3%; ride z=8 3.4% vs 0.0%; new z=7 14.9% vs 2.2%; tomorrow z=7 8.6% vs 0.7%; city z=7 5.7% vs 0.3%; atlanta z=7 2.3% vs 0.0%; chicago z=7 3.4% vs 0.1%; paris z=7 2.9% vs 0.1%; class z=7 2.3% vs 0.0%; vegas z=7 2.3% vs 0.0%; cheapest z=7 2.3% vs 0.0%; las z=6 2.3% vs 0.0%

Words, transport_traffic: traffic z=20 81.6% vs 0.0%; road z=11 6.6% vs 0.1%; is z=11 45.4% vs 9.4%; drive z=10 4.6% vs 0.0%; route z=10 4.6% vs 0.0%; home z=9 6.6% vs 0.2%; condition z=9 3.9% vs 0.0%; there z=9 15.8% vs 1.7%; situation z=8 3.3% vs 0.0%; heavy z=8 3.3% vs 0.0%; now z=8 12.5% vs 1.3%; right z=7 8.6% vs 0.7%; any z=7 11.2% vs 1.4%; level z=7 2.6% vs 0.0%; bad z=7 3.3% vs 0.1%; street z=7 3.3% vs 0.1%; highway z=6 2.0% vs 0.0%; current z=6 11.8% vs 1.8%; carmen z=6 2.0% vs 0.0%; work z=6 5.9% vs 0.4%

Words, weather_query: weather z=48 39.9% vs 1.6%; today z=30 19.1% vs 1.2%; is z=25 37.5% vs 9.1%; week z=23 10.1% vs 0.5%; rain z=23 8.5% vs 0.1%; be z=20 15.2% vs 2.3%; going z=19 9.1% vs 0.7%; will z=19 15.4% vs 2.5%; temperature z=19 6.1% vs 0.0%; should z=18 5.6% vs 0.2%; it z=17 21.1% vs 5.5%; tomorrow z=15 6.6% vs 0.7%; snow z=15 3.3% vs 0.0%; sunny z=14 3.3% vs 0.0%; outside z=14 4.9% vs 0.0%; cold z=12 2.5% vs 0.1%; forecast z=12 5.9% vs 1.0%; in z=11 22.2% vs 9.8%; bring z=11 2.0% vs 0.1%; hot z=11 2.5% vs 0.2%

2–4-word phrases, add_to_playlist: to my z=67 41.2% vs 0.5%; to the z=34 12.5% vs 1.7%; add this z=29 13.4% vs 0.0%; this is z=27 5.5% vs 0.1%; my playlist z=25 7.5% vs 0.0%; add a z=24 4.4% vs 0.1%; song to z=21 5.7% vs 0.0%; this song z=21 3.4% vs 0.2%; add the z=21 8.8% vs 0.0%; the playlist z=20 3.3% vs 0.1%; to add z=20 2.9% vs 0.1%; this album z=18 2.4% vs 0.1%; the artist z=16 2.1% vs 0.0%; song to my z=15 2.9% vs 0.0%; please add z=15 1.7% vs 0.1%; this song to z=14 2.1% vs 0.0%; put this z=14 4.2% vs 0.0%; you add z=14 1.6% vs 0.0%; in my z=14 2.6% vs 0.4%; can you add z=14 1.6% vs 0.0%

2–4-word phrases, alarm_query: set for z=19 18.5% vs 0.0%; alarm set z=15 11.6% vs 0.0%; my alarm z=14 8.9% vs 0.0%; the alarm z=14 8.9% vs 0.1%; my alarms z=13 11.0% vs 0.0%; for tomorrow z=13 9.6% vs 0.2%; have set z=13 11.6% vs 0.0%; alarm set for z=12 7.5% vs 0.0%; do i have z=12 15.1% vs 0.9%; i have set z=11 8.2% vs 0.0%; alarms set z=11 10.3% vs 0.0%; for tomorrow morning z=10 4.8% vs 0.0%; the morning z=10 5.5% vs 0.1%; tomorrow morning z=10 5.5% vs 0.1%; any alarms z=10 11.0% vs 0.0%; is my z=10 6.8% vs 0.2%; i set z=10 4.8% vs 0.1%; set for tomorrow z=10 5.5% vs 0.0%; you set z=9 4.1% vs 0.0%; set for today z=9 4.1% vs 0.0%

2–4-word phrases, alarm_remove: the alarm z=17 17.4% vs 0.1%; alarm for z=12 12.0% vs 0.2%; please remove z=12 8.7% vs 0.1%; my alarm z=11 7.6% vs 0.1%; remove the z=11 8.7% vs 0.1%; remove my z=10 7.6% vs 0.1%; please remove the z=10 5.4% vs 0.0%; alarm for tomorrow z=10 5.4% vs 0.0%; for tomorrow z=10 8.7% vs 0.2%; set for z=9 6.5% vs 0.1%; am alarm z=9 8.7% vs 0.0%; turn off z=9 8.7% vs 0.3%; is set z=9 4.3% vs 0.0%; my alarm for z=9 4.3% vs 0.0%; get rid of z=8 4.3% vs 0.0%; get rid z=8 4.3% vs 0.0%; rid of z=8 4.3% vs 0.0%; wake up z=8 4.3% vs 0.0%; my alarms z=8 4.3% vs 0.0%; delete the z=8 5.4% vs 0.1%

2–4-word phrases, alarm_set: alarm for z=26 38.5% vs 0.0%; set an z=20 20.4% vs 0.0%; an alarm z=19 33.5% vs 0.0%; please set z=19 11.8% vs 0.1%; me up z=18 10.0% vs 0.1%; at six z=16 8.6% vs 0.1%; six am z=16 9.0% vs 0.0%; at five z=16 8.1% vs 0.1%; seven am z=15 7.2% vs 0.0%; my alarm z=15 7.2% vs 0.0%; five am z=14 7.2% vs 0.0%; the alarm z=14 6.3% vs 0.1%; eight am z=14 6.3% vs 0.1%; alarm at z=14 10.4% vs 0.0%; set my z=14 6.8% vs 0.1%; am tomorrow z=14 6.8% vs 0.0%; set an alarm z=13 19.9% vs 0.0%; me at z=13 5.9% vs 0.1%; wake me z=13 10.4% vs 0.0%; up at z=13 10.0% vs 0.0%

2–4-word phrases, audio_volume_down: the volume z=19 30.0% vs 0.1%; lower the z=13 15.0% vs 0.0%; turn down z=11 10.0% vs 0.0%; the speakers z=11 10.0% vs 0.0%; volume to z=11 10.0% vs 0.0%; speaker volume z=10 8.3% vs 0.0%; turn the z=9 10.0% vs 0.1%; turn down the z=9 6.7% vs 0.0%; the volume to z=9 6.7% vs 0.0%; please lower z=9 6.7% vs 0.0%; down the z=9 6.7% vs 0.0%; turn the volume z=8 5.0% vs 0.0%; please lower the z=8 5.0% vs 0.0%; volume of z=7 5.0% vs 0.0%; your volume z=7 5.0% vs 0.0%; the speaker volume z=6 3.3% vs 0.0%; volume please z=6 3.3% vs 0.0%; twenty percent z=6 3.3% vs 0.0%; to ten z=6 3.3% vs 0.0%; the speaker z=6 3.3% vs 0.0%

2–4-word phrases, audio_volume_mute: the speakers z=11 5.0% vs 0.0%; until i z=10 4.2% vs 0.0%; the sound z=9 3.4% vs 0.0%; please do z=8 3.4% vs 0.0%; the volume z=8 4.2% vs 0.1%; the speaker z=8 2.5% vs 0.0%; turn off z=7 5.0% vs 0.3%; five minutes z=7 2.5% vs 0.0%; for five minutes z=7 2.5% vs 0.0%; volume to z=7 2.5% vs 0.0%; set to z=6 1.7% vs 0.0%; twenty minutes z=6 1.7% vs 0.0%; my speakers z=6 1.7% vs 0.0%; i say z=6 1.7% vs 0.0%; pause the z=6 1.7% vs 0.0%; for twenty z=6 1.7% vs 0.0%; the volume to z=6 1.7% vs 0.0%; for five z=5 2.5% vs 0.1%; turn the z=5 2.5% vs 0.1%; down the z=5 1.7% vs 0.0%

2–4-word phrases, audio_volume_up: the volume z=24 33.6% vs 0.0%; turn up z=16 14.5% vs 0.0%; increase the z=15 11.8% vs 0.0%; volume to z=14 10.9% vs 0.0%; turn up the z=13 8.2% vs 0.0%; volume of z=12 12.7% vs 0.0%; music player z=11 6.4% vs 0.0%; the music z=11 7.3% vs 0.1%; the volume to z=11 6.4% vs 0.0%; turn the z=11 7.3% vs 0.1%; the volume of z=10 11.8% vs 0.0%; up the z=10 9.1% vs 0.3%; speaker volume z=9 4.5% vs 0.0%; the speaker z=9 4.5% vs 0.0%; raise the z=9 4.5% vs 0.0%; please increase z=9 5.5% vs 0.0%; turn the volume z=9 4.5% vs 0.0%; to fifty z=8 3.6% vs 0.0%; please increase the z=8 3.6% vs 0.0%; a little z=8 3.6% vs 0.0%

2–4-word phrases, book_restaurant: book a z=47 46.3% vs 0.2%; at a z=32 27.6% vs 0.1%; need a z=28 9.9% vs 0.2%; i need a z=26 8.6% vs 0.2%; for a z=21 8.5% vs 0.8%; me a z=21 7.5% vs 0.6%; restaurant in z=21 13.9% vs 0.0%; to book z=20 9.4% vs 0.0%; a restaurant z=19 23.8% vs 0.0%; book me z=19 4.5% vs 0.1%; book me a z=18 4.1% vs 0.1%; and i z=17 5.4% vs 0.0%; at the z=17 6.0% vs 0.7%; to eat z=16 5.1% vs 0.0%; a party z=16 6.9% vs 0.0%; to book a z=16 7.6% vs 0.0%; i'd like z=16 5.3% vs 0.6%; for me z=15 4.9% vs 0.6%; for one z=15 2.9% vs 0.0%; a restaurant in z=15 9.2% vs 0.0%

2–4-word phrases, calendar_query: do i have z=38 20.1% vs 0.7%; do i z=28 21.0% vs 2.8%; i have z=24 22.1% vs 4.8%; is my z=24 7.3% vs 0.1%; have any z=22 6.1% vs 0.1%; i have any z=21 6.0% vs 0.1%; do i have any z=21 6.0% vs 0.1%; my calendar z=19 6.5% vs 0.4%; when is z=18 4.9% vs 0.2%; this week z=17 4.7% vs 0.2%; for today z=15 3.5% vs 0.2%; am i z=15 2.9% vs 0.1%; my schedule z=15 3.8% vs 0.0%; on my calendar z=14 2.7% vs 0.1%; what are z=14 4.3% vs 0.5%; what time z=13 5.5% vs 1.0%; are my z=13 2.2% vs 0.1%; the event z=13 2.1% vs 0.0%; scheduled for z=13 2.2% vs 0.0%; my next z=12 2.1% vs 0.1%

2–4-word phrases, calendar_remove: my calendar z=36 23.9% vs 0.3%; from my z=21 9.1% vs 0.2%; the calendar z=19 7.1% vs 0.1%; on my calendar z=17 5.3% vs 0.1%; delete my z=17 5.0% vs 0.1%; next event z=16 6.8% vs 0.0%; calendar for z=16 4.8% vs 0.1%; all the z=16 5.0% vs 0.1%; remove the z=16 4.8% vs 0.1%; all my z=15 4.3% vs 0.0%; delete the z=15 4.3% vs 0.1%; remove all z=15 6.3% vs 0.0%; all events z=14 9.6% vs 0.0%; please delete z=14 4.0% vs 0.0%; please remove z=14 3.5% vs 0.1%; the events z=14 3.5% vs 0.0%; my next z=14 3.5% vs 0.1%; my calendar for z=13 3.0% vs 0.1%; event on z=12 2.5% vs 0.0%; delete all z=11 9.8% vs 0.0%

2–4-word phrases, calendar_set: my calendar z=29 9.6% vs 0.3%; remind me z=28 21.6% vs 0.0%; set a z=25 7.8% vs 0.1%; meeting with z=23 9.4% vs 0.0%; a meeting z=21 8.7% vs 0.0%; me about z=18 5.3% vs 0.4%; two pm z=17 3.1% vs 0.1%; at two z=16 3.0% vs 0.1%; in my calendar z=16 3.0% vs 0.0%; calendar for z=16 2.9% vs 0.1%; three pm z=16 2.8% vs 0.0%; at ten z=15 2.7% vs 0.1%; reminder for z=15 6.3% vs 0.0%; remind me to z=15 6.0% vs 0.0%; the meeting z=15 2.6% vs 0.1%; me to z=15 6.6% vs 1.3%; my meeting z=15 2.6% vs 0.1%; tomorrow at z=15 2.5% vs 0.0%; on tuesday z=15 3.0% vs 0.0%; at two pm z=15 2.5% vs 0.0%

2–4-word phrases, cancel_order: to cancel z=33 24.6% vs 0.3%; purchase ord z=21 11.6% vs 0.3%; order ord z=21 12.1% vs 0.3%; i cancel z=16 5.6% vs 0.1%; cancel it z=15 5.6% vs 0.1%; help with z=15 5.1% vs 0.1%; me cancel z=13 4.2% vs 0.0%; need help with z=13 4.0% vs 0.0%; help canceling z=13 4.0% vs 0.0%; how to cancel z=13 4.2% vs 0.0%; help me cancel z=13 4.0% vs 0.0%; no longer z=12 3.4% vs 0.0%; i bought z=12 4.2% vs 0.0%; help cancelling z=12 3.4% vs 0.0%; assistance to cancel z=12 3.4% vs 0.0%; do to cancel z=12 3.4% vs 0.1%; to do to cancel z=12 3.4% vs 0.1%; assistance canceling z=12 3.4% vs 0.0%; help to cancel z=12 3.1% vs 0.0%; assistance cancelling z=12 3.1% vs 0.0%

2–4-word phrases, change_order: of order z=32 17.4% vs 0.2%; order ord z=29 13.6% vs 0.2%; of purchase z=28 13.3% vs 0.2%; purchase ord z=25 10.2% vs 0.2%; an item z=24 9.7% vs 0.1%; to order z=22 8.3% vs 0.2%; an article z=20 6.6% vs 0.1%; to purchase z=20 6.5% vs 0.1%; some items z=19 5.9% vs 0.1%; a product z=18 7.7% vs 0.4%; to add z=18 5.8% vs 0.1%; some products z=17 5.0% vs 0.1%; of order ord z=16 4.3% vs 0.1%; several items z=15 4.0% vs 0.1%; of purchase ord z=15 3.8% vs 0.1%; something from z=15 3.7% vs 0.0%; some articles z=15 3.7% vs 0.1%; to delete z=14 3.8% vs 0.1%; to change z=14 5.3% vs 0.4%; item to z=14 3.1% vs 0.0%

2–4-word phrases, change_shipping_address: shipping address z=39 29.1% vs 0.8%; delivery address z=38 26.7% vs 0.7%; trying to z=32 20.3% vs 0.6%; my delivery z=28 13.6% vs 0.2%; the shipping z=25 11.5% vs 0.2%; my shipping z=25 12.1% vs 0.1%; my shipping address z=25 11.9% vs 0.1%; my delivery address z=25 13.2% vs 0.1%; the shipping address z=24 11.1% vs 0.1%; to modify z=22 8.4% vs 0.1%; the delivery z=21 8.4% vs 0.2%; to update z=21 8.0% vs 0.1%; the delivery address z=21 7.8% vs 0.1%; to correct z=20 7.5% vs 0.1%; to edit z=20 7.2% vs 0.1%; to change z=19 8.3% vs 0.4%; the address z=19 15.0% vs 0.0%; change my z=17 5.4% vs 0.0%; update my z=17 5.5% vs 0.0%; edit my z=16 4.6% vs 0.1%

2–4-word phrases, check_cancellation_fee: see the z=40 34.7% vs 1.0%; check the z=39 31.5% vs 0.8%; to see the z=34 24.2% vs 0.7%; to check the z=33 21.5% vs 0.4%; the termination z=25 16.4% vs 0.1%; the cancellation z=24 14.7% vs 0.1%; to check z=24 21.7% vs 1.7%; to see z=23 25.0% vs 2.5%; seeing the z=20 8.5% vs 0.2%; checking the z=19 7.4% vs 0.1%; i check the z=16 5.4% vs 0.1%; do to see the z=14 3.9% vs 0.1%; assistance checking the z=14 3.9% vs 0.0%; help seeing the z=14 3.8% vs 0.1%; i see the z=14 4.4% vs 0.2%; do to see z=13 4.0% vs 0.1%; to do to see z=13 3.9% vs 0.1%; do to check the z=13 3.2% vs 0.0%; assistance checking z=12 3.9% vs 0.2%; do to check z=12 3.2% vs 0.1%

2–4-word phrases, check_invoice: my bill z=26 13.6% vs 0.1%; my invoice z=25 12.2% vs 0.2%; bill from z=24 11.5% vs 0.1%; the bill z=24 11.5% vs 0.2%; the invoice z=23 10.5% vs 0.2%; look for z=23 10.5% vs 0.1%; bills from z=23 10.1% vs 0.1%; invoices from z=22 9.2% vs 0.1%; bill 00108 z=20 8.2% vs 0.1%; give a z=20 8.9% vs 0.1%; invoice from z=20 7.9% vs 0.2%; bill 12588 z=19 7.5% vs 0.1%; bill 37777 z=19 7.4% vs 0.1%; to locate z=19 7.6% vs 0.1%; at my z=19 7.8% vs 0.0%; invoice 12588 z=18 6.8% vs 0.1%; invoice 85632 z=18 6.7% vs 0.1%; to give z=18 6.7% vs 0.1%; invoice 37777 z=18 6.5% vs 0.1%; see my z=18 6.2% vs 0.1%

2–4-word phrases, check_payment_methods: see your z=20 7.4% vs 0.1%; list the z=19 10.7% vs 0.0%; to see z=19 20.3% vs 2.6%; the payment z=19 11.6% vs 0.0%; the available z=17 8.3% vs 0.0%; to check z=17 15.0% vs 1.8%; to see your z=17 5.4% vs 0.1%; check your z=17 5.3% vs 0.1%; options are z=16 4.9% vs 0.1%; to list z=16 13.7% vs 0.0%; see what z=15 6.7% vs 0.4%; to check your z=15 4.1% vs 0.0%; see the z=15 11.2% vs 1.3%; methods are z=14 4.1% vs 0.1%; see ur z=14 3.7% vs 0.0%; to see what z=13 4.7% vs 0.3%; check the z=13 9.0% vs 1.1%; to see the z=12 7.8% vs 0.9%; check ur z=12 3.0% vs 0.0%; to check the z=12 6.3% vs 0.6%

2–4-word phrases, check_refund_policy: can i z=40 64.7% vs 3.8%; ask for z=32 24.9% vs 0.2%; money back z=24 17.2% vs 0.1%; a refund z=24 13.1% vs 0.2%; i ask z=24 34.4% vs 0.0%; to see z=22 28.4% vs 2.4%; to be z=21 13.5% vs 0.5%; request a z=20 9.8% vs 0.1%; ask for a z=19 9.1% vs 0.1%; i ask for z=19 24.7% vs 0.0%; can i ask z=19 34.1% vs 0.0%; can i ask for z=17 24.6% vs 0.0%; see your z=17 7.1% vs 0.1%; i request z=17 28.2% vs 0.0%; how long z=17 10.3% vs 0.5%; my money back z=17 6.7% vs 0.1%; to check z=17 18.3% vs 1.7%; to see your z=16 6.1% vs 0.1%; in what z=15 28.2% vs 0.0%; check your z=15 5.3% vs 0.1%

2–4-word phrases, complaint: make a z=35 25.6% vs 0.2%; to make z=29 18.4% vs 0.1%; your company z=25 13.3% vs 0.3%; a complaint z=25 13.5% vs 0.1%; to make a z=23 17.3% vs 0.0%; ur company z=21 9.0% vs 0.2%; i make z=14 4.2% vs 0.0%; a consumer z=12 22.2% vs 0.0%; a customer z=12 21.8% vs 0.0%; making a z=11 4.6% vs 0.0%; i make a z=11 4.2% vs 0.0%; make a complaint z=10 3.8% vs 0.0%; do i make z=9 1.5% vs 0.0%; can i make z=8 2.0% vs 0.0%; customer complaint z=8 8.0% vs 0.0%; help making z=8 2.5% vs 0.0%; consumer complaint z=8 7.6% vs 0.0%; how to make z=8 1.3% vs 0.0%; experience with z=8 2.4% vs 0.0%; a customer complaint z=8 6.6% vs 0.0%

2–4-word phrases, contact_customer_service: i can z=40 37.8% vs 1.1%; customer service z=30 32.0% vs 0.1%; what time z=28 20.3% vs 0.7%; to talk z=22 11.2% vs 0.3%; at what z=20 19.5% vs 0.0%; see what z=19 9.8% vs 0.4%; to speak z=19 8.8% vs 0.3%; talk to z=19 8.6% vs 0.2%; to talk to z=18 6.9% vs 0.1%; at what time z=17 19.4% vs 0.0%; check what z=17 7.8% vs 0.3%; time i z=16 15.5% vs 0.0%; to call z=16 6.1% vs 0.0%; speak to z=15 5.3% vs 0.2%; to customer z=15 14.1% vs 0.0%; talk with z=14 5.3% vs 0.2%; to contact z=14 4.8% vs 0.1%; speak with z=14 5.2% vs 0.2%; to talk with z=13 4.3% vs 0.1%; to speak with z=13 4.5% vs 0.1%

2–4-word phrases, contact_human_agent: with a z=31 20.2% vs 0.5%; to talk z=30 16.7% vs 0.2%; to speak z=29 17.1% vs 0.1%; talk to z=26 13.2% vs 0.2%; to a z=25 12.0% vs 0.1%; speak with z=25 13.4% vs 0.1%; talk with z=24 11.5% vs 0.1%; speak to z=23 10.8% vs 0.1%; with an z=22 17.1% vs 0.0%; to talk to z=22 9.1% vs 0.1%; to speak with z=22 9.5% vs 0.1%; to contact z=21 8.7% vs 0.1%; to talk with z=20 7.5% vs 0.1%; to speak to z=20 7.5% vs 0.1%; to an z=16 8.7% vs 0.0%; do to talk z=14 4.1% vs 0.0%; to do to talk z=14 4.1% vs 0.0%; do to speak z=14 4.1% vs 0.0%; to do to z=14 14.2% vs 2.4%; to do to speak z=14 4.1% vs 0.0%

2–4-word phrases, cooking_recipe: how do z=17 13.2% vs 0.9%; me how z=15 5.2% vs 0.1%; how do you z=14 4.5% vs 0.0%; me how to z=13 4.9% vs 0.0%; do you z=13 6.3% vs 0.3%; to make z=13 6.3% vs 0.4%; how do i z=12 8.7% vs 0.9%; do i z=12 14.3% vs 3.0%; tell me how z=12 3.5% vs 0.0%; how to z=12 11.8% vs 2.0%; the recipe z=12 6.6% vs 0.0%; tell me how to z=12 3.5% vs 0.0%; how to make z=12 3.1% vs 0.0%; do i make z=12 3.1% vs 0.0%; how do i make z=11 3.1% vs 0.0%; i make z=11 3.5% vs 0.1%; should i z=10 3.1% vs 0.1%; recipe for z=9 15.7% vs 0.0%; how long z=9 5.2% vs 0.6%; the best z=9 3.5% vs 0.2%

2–4-word phrases, create_account: create a z=33 25.7% vs 0.1%; for my z=29 18.3% vs 0.5%; standard account z=26 15.4% vs 0.5%; open a z=25 21.0% vs 0.0%; a standard z=22 11.7% vs 0.1%; a standard account z=22 11.4% vs 0.1%; pro account z=22 10.5% vs 0.3%; platinum account z=21 10.0% vs 0.3%; premium account z=19 8.6% vs 0.3%; freemium account z=19 8.1% vs 0.3%; a pro z=19 7.7% vs 0.0%; gold account z=18 8.1% vs 0.3%; a pro account z=18 7.5% vs 0.0%; a premium z=17 6.4% vs 0.0%; a platinum z=17 6.6% vs 0.0%; a platinum account z=17 6.4% vs 0.0%; a premium account z=17 5.9% vs 0.0%; a gold z=16 5.5% vs 0.0%; a gold account z=16 5.3% vs 0.0%; a freemium z=16 5.6% vs 0.0%

2–4-word phrases, datetime_convert: the time z=18 32.8% vs 0.2%; time zone z=16 22.4% vs 0.0%; what is the time z=14 16.4% vs 0.1%; is the time z=14 16.4% vs 0.1%; difference between z=12 16.4% vs 0.0%; what time is it z=11 13.4% vs 0.1%; time is it z=11 13.4% vs 0.1%; time is it in z=11 11.9% vs 0.1%; time is z=11 19.4% vs 0.4%; is it in z=10 11.9% vs 0.1%; it in z=10 11.9% vs 0.1%; new york z=10 14.9% vs 0.3%; what time is z=10 16.4% vs 0.4%; what time z=9 22.4% vs 1.0%; time to z=9 7.5% vs 0.0%; g m z=9 7.5% vs 0.0%; g m t z=9 7.5% vs 0.0%; m t z=9 7.5% vs 0.0%; eastern time z=8 6.0% vs 0.0%; the time zone z=8 7.5% vs 0.0%

2–4-word phrases, datetime_query: the time z=30 21.8% vs 0.1%; what time z=27 18.7% vs 0.9%; time in z=26 20.7% vs 0.0%; time is z=26 14.4% vs 0.3%; what time is z=25 13.3% vs 0.3%; is it z=23 17.9% vs 1.3%; it in z=23 10.7% vs 0.1%; is it in z=22 10.7% vs 0.1%; is the z=19 19.4% vs 2.7%; time is it z=19 13.1% vs 0.0%; what time is it z=18 12.4% vs 0.0%; what is z=18 18.1% vs 2.6%; is the time z=17 7.4% vs 0.0%; the week z=17 6.1% vs 0.0%; what is the z=17 14.8% vs 1.9%; what is the time z=17 7.0% vs 0.0%; what day z=17 12.0% vs 0.0%; time is it in z=17 9.8% vs 0.0%; tell me z=17 13.7% vs 1.7%; the date z=17 7.4% vs 0.0%

2–4-word phrases, delete_account: standard account z=28 17.3% vs 0.5%; information about z=24 13.7% vs 0.5%; platinum account z=22 10.6% vs 0.3%; about the z=21 11.3% vs 0.5%; to delete z=21 9.1% vs 0.1%; gold account z=20 9.5% vs 0.3%; premium account z=20 9.3% vs 0.3%; pro account z=20 9.6% vs 0.3%; information about the z=19 7.4% vs 0.1%; to remove z=19 7.6% vs 0.1%; to cancel z=19 8.6% vs 0.3%; to do to z=17 17.7% vs 2.4%; do to z=17 17.7% vs 2.4%; with the z=17 6.4% vs 0.2%; freemium account z=17 7.1% vs 0.3%; know what z=16 13.6% vs 1.6%; problem with z=15 6.8% vs 0.4%; to do z=15 17.8% vs 2.9%; know what i z=15 9.3% vs 0.9%; of a z=14 7.6% vs 0.6%

2–4-word phrases, delivery_options: see what z=31 25.1% vs 0.3%; check what z=27 18.5% vs 0.2%; to see what z=26 17.3% vs 0.2%; to check what z=23 13.4% vs 0.1%; to see z=19 27.5% vs 2.6%; are there z=19 11.0% vs 0.3%; methods are z=19 9.2% vs 0.1%; are available z=18 8.3% vs 0.1%; to check z=18 21.0% vs 1.8%; options are z=16 6.3% vs 0.1%; options i z=15 10.8% vs 0.0%; checking what z=15 5.9% vs 0.0%; for delivery z=15 13.6% vs 0.0%; the shipment z=15 5.5% vs 0.1%; i see what z=13 4.7% vs 0.1%; methods are available z=13 4.7% vs 0.0%; methods i z=13 10.6% vs 0.0%; options you z=13 4.1% vs 0.0%; seeing what z=12 4.1% vs 0.1%; to see the z=12 10.2% vs 0.9%

2–4-word phrases, delivery_period: i expect z=33 36.5% vs 0.1%; how long z=30 22.0% vs 0.3%; see how z=27 25.3% vs 0.1%; can i z=26 44.2% vs 4.1%; check how z=24 21.5% vs 0.1%; to see how z=24 18.0% vs 0.1%; when will z=23 15.6% vs 0.1%; to see z=22 28.9% vs 2.4%; see how long z=20 9.8% vs 0.1%; is going z=20 10.2% vs 0.1%; to check how z=20 12.9% vs 0.0%; to check z=19 20.4% vs 1.7%; check how long z=18 8.4% vs 0.1%; when my z=18 15.8% vs 0.0%; will my z=17 15.8% vs 0.0%; going to z=17 10.3% vs 0.5%; to see how long z=17 6.7% vs 0.1%; is going to z=15 10.2% vs 0.0%; i check z=15 8.1% vs 0.4%; when will my z=14 15.3% vs 0.0%

2–4-word phrases, edit_account: on my z=26 15.1% vs 0.7%; my user z=23 12.9% vs 0.6%; information on z=22 13.8% vs 0.0%; to edit z=19 6.9% vs 0.2%; standard account z=19 9.6% vs 0.6%; premium account z=19 7.9% vs 0.3%; freemium account z=18 7.4% vs 0.3%; platinum account z=18 7.7% vs 0.4%; edit my z=18 5.5% vs 0.1%; gold account z=17 7.1% vs 0.3%; pro account z=16 6.9% vs 0.4%; details on z=16 15.6% vs 0.0%; to edit my z=16 4.5% vs 0.0%; info on z=15 16.3% vs 0.0%; to modify z=15 4.5% vs 0.2%; correct the z=14 3.5% vs 0.1%; modify the z=14 3.5% vs 0.1%; the details z=14 3.5% vs 0.1%; update the z=14 3.4% vs 0.1%; to update z=13 3.9% vs 0.2%

2–4-word phrases, email_addcontact: dot com z=19 28.4% vs 0.1%; my contacts z=18 29.9% vs 0.0%; gmail dot z=14 16.4% vs 0.1%; gmail dot com z=14 16.4% vs 0.1%; new email z=13 17.9% vs 0.2%; this email z=13 14.9% vs 0.0%; email to z=13 22.4% vs 0.4%; my contact z=13 13.4% vs 0.0%; email to my z=12 13.4% vs 0.1%; new contact z=11 13.4% vs 0.0%; a new z=10 16.4% vs 0.3%; to my z=10 31.3% vs 1.8%; add a new z=10 9.0% vs 0.0%; email in z=10 10.4% vs 0.0%; to my contacts z=10 19.4% vs 0.0%; email address z=10 9.0% vs 0.0%; in my contacts z=10 7.5% vs 0.0%; dot com to z=9 13.4% vs 0.0%; com to z=9 13.4% vs 0.0%; a new email z=9 7.5% vs 0.0%

2–4-word phrases, email_query: any new z=23 13.3% vs 0.0%; new email z=23 8.1% vs 0.1%; have any z=21 7.2% vs 0.2%; i have any z=21 6.8% vs 0.1%; do i have any z=19 5.8% vs 0.1%; do i have z=19 9.6% vs 0.9%; did i z=18 5.3% vs 0.1%; have i z=18 5.4% vs 0.0%; there any z=15 4.7% vs 0.2%; the last z=15 5.6% vs 0.4%; are there any z=14 3.7% vs 0.1%; are there z=14 4.6% vs 0.3%; email from z=14 8.2% vs 0.0%; from my z=14 4.2% vs 0.3%; check my z=13 3.7% vs 0.2%; if i z=13 2.8% vs 0.1%; i got z=13 2.6% vs 0.0%; my email z=12 3.7% vs 0.0%; please check z=12 3.2% vs 0.2%; the email z=12 2.3% vs 0.1%

2–4-word phrases, email_querycontact: phone number z=16 11.5% vs 0.0%; my contacts z=13 6.4% vs 0.1%; in my contacts z=11 5.1% vs 0.0%; the contact z=11 6.4% vs 0.0%; number of z=10 3.8% vs 0.0%; john smith z=10 3.8% vs 0.0%; contact information z=10 4.5% vs 0.0%; in my z=10 7.7% vs 0.5%; what is z=9 14.7% vs 2.7%; number for z=9 3.8% vs 0.0%; address of z=9 5.1% vs 0.0%; a contact z=9 3.2% vs 0.0%; my phone z=8 2.6% vs 0.0%; and tell me z=8 2.6% vs 0.0%; open the z=8 2.6% vs 0.0%; address for z=8 2.6% vs 0.0%; of john z=8 2.6% vs 0.0%; and tell z=7 2.6% vs 0.1%; email address z=7 2.6% vs 0.1%; tell me z=7 9.0% vs 1.8%

2–4-word phrases, email_sendemail: email to z=30 39.0% vs 0.0%; an email z=30 23.4% vs 0.1%; send an z=22 15.2% vs 0.0%; dot com z=20 6.9% vs 0.1%; new email z=20 6.5% vs 0.1%; email to my z=17 7.5% vs 0.0%; send a z=17 5.1% vs 0.1%; the email z=17 5.0% vs 0.0%; to my z=16 11.5% vs 1.8%; i will z=16 4.2% vs 0.0%; gmail dot com z=16 5.0% vs 0.0%; gmail dot z=16 5.0% vs 0.0%; mail to z=15 3.6% vs 0.1%; email address z=14 3.8% vs 0.0%; please send z=14 5.5% vs 0.0%; that i z=13 3.4% vs 0.1%; will be z=13 3.8% vs 0.2%; an email to z=12 17.6% vs 0.0%; my friend z=12 2.6% vs 0.0%; email for z=12 2.6% vs 0.0%

2–4-word phrases, general_affirm: you have z=24 11.9% vs 0.5%; the command z=23 10.5% vs 0.0%; response to z=17 10.7% vs 0.0%; the answer z=14 3.3% vs 0.1%; you've done z=13 3.5% vs 0.1%; got the z=13 3.3% vs 0.0%; you've understood z=13 4.0% vs 0.0%; you've got z=13 4.7% vs 0.0%; you did z=12 3.5% vs 0.2%; you have got z=12 3.7% vs 0.0%; you got z=12 2.6% vs 0.0%; correct response z=12 2.3% vs 0.0%; you have done z=12 3.0% vs 0.2%; have done z=12 3.0% vs 0.2%; you understood z=11 2.6% vs 0.0%; have gotten z=11 2.1% vs 0.0%; you have gotten z=11 2.1% vs 0.0%; get it z=11 1.9% vs 0.0%; answer was z=11 1.9% vs 0.0%; have understood z=11 3.0% vs 0.0%

2–4-word phrases, general_commandstop: you just z=23 15.1% vs 0.1%; cancel it z=21 12.4% vs 0.1%; would you z=19 12.7% vs 0.3%; will you z=18 13.1% vs 0.5%; that is z=17 11.2% vs 0.3%; shut off z=15 10.8% vs 0.0%; like you z=15 7.6% vs 0.2%; olly please z=13 8.0% vs 0.0%; shut down z=13 12.7% vs 0.0%; you cancel z=13 8.0% vs 0.0%; could you z=12 10.4% vs 0.9%; you shut z=12 24.3% vs 0.0%; it please z=12 4.4% vs 0.1%; please can you z=12 4.4% vs 0.1%; please can z=12 4.4% vs 0.1%; please shut z=12 8.0% vs 0.0%; to you z=12 4.8% vs 0.0%; please would you z=11 3.6% vs 0.1%; please would z=11 3.6% vs 0.1%; you please z=11 6.8% vs 0.5%

2–4-word phrases, general_confirm: like you z=24 16.4% vs 0.1%; please check z=24 13.0% vs 0.1%; you to z=23 12.1% vs 0.1%; said please z=22 14.2% vs 0.0%; my last z=22 13.2% vs 0.1%; you check z=22 33.1% vs 0.0%; you please z=22 13.5% vs 0.4%; would you z=20 11.4% vs 0.3%; will you z=20 12.6% vs 0.5%; check and z=18 30.8% vs 0.0%; i said z=18 9.8% vs 0.0%; could you z=17 13.0% vs 0.8%; the details z=17 6.4% vs 0.1%; i want you z=16 5.7% vs 0.1%; like you to z=16 8.2% vs 0.0%; want you z=16 5.7% vs 0.1%; i like z=15 5.5% vs 0.1%; did you z=15 5.3% vs 0.0%; you get z=15 5.3% vs 0.1%; i'd like you z=14 6.4% vs 0.0%

2–4-word phrases, general_dontcare: to me z=22 14.5% vs 0.4%; with me z=21 10.7% vs 0.1%; for me z=15 10.1% vs 0.7%; would be z=14 4.9% vs 0.0%; for any z=14 5.2% vs 0.0%; one olly z=13 5.5% vs 0.0%; do not z=12 12.5% vs 2.0%; will be z=11 4.3% vs 0.2%; that is z=10 4.6% vs 0.3%; it is z=10 3.8% vs 0.2%; is of z=10 5.8% vs 0.0%; be good z=10 2.6% vs 0.0%; is good z=9 2.0% vs 0.0%; one is z=9 2.0% vs 0.0%; i would z=9 5.8% vs 0.8%; about what z=8 2.3% vs 0.1%; mind to z=8 8.4% vs 0.0%; ok to z=8 1.4% vs 0.0%; is ok z=8 2.0% vs 0.0%; for what z=8 2.0% vs 0.0%

2–4-word phrases, general_explain: to me z=28 16.7% vs 0.3%; me on z=23 12.1% vs 0.1%; what you z=23 16.7% vs 0.0%; once again z=22 10.3% vs 0.1%; just said z=22 10.3% vs 0.2%; you please z=20 10.3% vs 0.4%; me about z=19 10.3% vs 0.4%; was not z=18 7.0% vs 0.1%; i was z=18 6.8% vs 0.1%; about your z=18 7.5% vs 0.2%; what you've z=17 5.9% vs 0.0%; about what z=16 7.3% vs 0.0%; me more z=16 5.5% vs 0.1%; your answer z=16 7.0% vs 0.0%; you have z=16 8.3% vs 0.5%; what you have z=16 8.3% vs 0.0%; will you z=15 7.9% vs 0.5%; your answers z=15 10.3% vs 0.0%; up with z=15 4.6% vs 0.1%; with you z=15 4.6% vs 0.1%

2–4-word phrases, general_joke: tell me a z=17 19.6% vs 0.0%; tell me z=13 28.3% vs 1.8%; me a z=13 20.7% vs 0.8%; a joke z=12 26.1% vs 0.0%; a good z=11 8.7% vs 0.1%; hear a z=9 5.4% vs 0.1%; me a joke z=8 13.0% vs 0.0%; tell me a joke z=8 12.0% vs 0.0%; you know z=8 4.3% vs 0.0%; to hear a z=8 4.3% vs 0.0%; make me z=8 5.4% vs 0.1%; a funny z=7 7.6% vs 0.0%; me a good z=7 3.3% vs 0.0%; tell me a good z=7 3.3% vs 0.0%; you tell me a z=7 3.3% vs 0.0%; do you know z=7 3.3% vs 0.0%; want to hear a z=7 3.3% vs 0.0%; hey olly z=7 3.3% vs 0.0%; know any z=6 2.2% vs 0.0%; tell me some z=6 2.2% vs 0.0%

2–4-word phrases, general_negate: i think z=30 33.0% vs 0.0%; what i z=23 14.5% vs 1.2%; is not z=22 8.6% vs 0.1%; i am z=22 12.8% vs 1.0%; was not z=22 8.2% vs 0.1%; you did z=22 7.7% vs 0.1%; you have z=20 9.0% vs 0.5%; did not z=18 6.8% vs 0.0%; it was z=14 3.4% vs 0.1%; not get z=13 3.0% vs 0.0%; i did z=13 3.0% vs 0.0%; i did not z=13 3.0% vs 0.0%; right response z=13 3.5% vs 0.0%; i was z=13 3.0% vs 0.1%; up with z=13 2.7% vs 0.1%; correct response z=12 2.9% vs 0.0%; i didn't z=12 2.6% vs 0.0%; the correct z=12 8.0% vs 0.0%; it is z=12 3.3% vs 0.2%; have done z=12 3.0% vs 0.1%

2–4-word phrases, general_praise: you are z=24 9.5% vs 0.1%; you did z=24 8.3% vs 0.1%; have done z=24 8.8% vs 0.1%; you have done z=23 8.8% vs 0.1%; you've done z=23 8.2% vs 0.1%; you have z=21 8.8% vs 0.5%; thank you z=21 29.2% vs 0.0%; that was z=19 5.5% vs 0.0%; it was z=19 5.1% vs 0.1%; done it z=18 6.3% vs 0.0%; done that z=16 5.9% vs 0.0%; are my z=14 2.9% vs 0.1%; it right z=14 3.0% vs 0.0%; it is z=14 3.5% vs 0.2%; have done it z=12 2.9% vs 0.0%; that right z=12 2.9% vs 0.0%; you have done it z=12 2.9% vs 0.0%; that is z=12 3.9% vs 0.3%; have done that z=12 3.5% vs 0.0%; so much z=12 10.4% vs 0.0%

2–4-word phrases, general_quirky: do you z=20 5.1% vs 0.3%; tell me z=17 7.6% vs 1.7%; are you z=15 2.5% vs 0.0%; what do you z=13 1.9% vs 0.0%; i had z=12 2.0% vs 0.0%; i would z=11 3.3% vs 0.8%; i would like z=11 2.9% vs 0.6%; tell me about z=11 2.0% vs 0.2%; would like z=11 2.9% vs 0.6%; the best z=11 1.9% vs 0.2%; my day z=10 2.0% vs 0.0%; would be z=10 1.3% vs 0.1%; what would z=10 1.4% vs 0.0%; i got z=10 1.1% vs 0.1%; if you z=10 1.0% vs 0.0%; me how z=10 1.3% vs 0.1%; me about z=10 2.4% vs 0.5%; day today z=10 1.0% vs 0.0%; it to z=10 1.3% vs 0.1%; tell me how z=9 1.0% vs 0.0%

2–4-word phrases, general_repeat: to hear z=25 15.1% vs 0.4%; once again z=22 10.5% vs 0.1%; you tell me z=22 11.2% vs 0.3%; you tell z=22 11.2% vs 0.3%; just said z=21 9.6% vs 0.2%; would you z=21 10.5% vs 0.3%; the last z=20 11.2% vs 0.4%; tell me z=20 18.6% vs 1.7%; like to hear z=20 8.8% vs 0.1%; again please z=19 19.5% vs 0.0%; will you z=18 9.9% vs 0.5%; i'd like to hear z=16 5.3% vs 0.0%; what you've z=15 4.8% vs 0.1%; one more z=15 29.2% vs 0.0%; you say z=14 7.7% vs 0.0%; it again z=14 5.7% vs 0.0%; could you z=14 9.2% vs 0.9%; to hear the z=14 4.2% vs 0.1%; more time z=14 19.5% vs 0.0%; hear the z=13 4.2% vs 0.1%

2–4-word phrases, get_invoice: to get z=35 25.4% vs 0.4%; my invoice z=27 15.3% vs 0.2%; the bill z=27 15.3% vs 0.2%; get the z=26 13.8% vs 0.2%; the invoice z=25 12.7% vs 0.2%; invoice from z=23 11.2% vs 0.1%; to do to z=23 25.1% vs 2.3%; do to z=23 25.2% vs 2.3%; know what i z=23 15.7% vs 0.8%; bill from z=22 10.1% vs 0.2%; know what z=22 19.7% vs 1.6%; to do z=21 25.8% vs 2.8%; my bill z=20 9.2% vs 0.2%; invoices from z=20 8.6% vs 0.1%; what i z=20 16.0% vs 1.2%; bills from z=20 8.5% vs 0.1%; invoice 12588 z=19 7.7% vs 0.1%; invoice 00108 z=19 7.4% vs 0.1%; do to get z=19 11.5% vs 0.0%; bill 37777 z=18 7.1% vs 0.1%

2–4-word phrases, get_refund: my money z=35 32.4% vs 0.1%; of my z=34 27.9% vs 1.1%; a refund z=23 10.4% vs 0.2%; get a z=22 10.1% vs 0.1%; a reimbursement z=22 9.1% vs 0.1%; ask for z=22 11.1% vs 0.4%; to receive z=20 7.7% vs 0.1%; a rebate z=20 13.3% vs 0.0%; to get z=18 9.8% vs 0.6%; a compensation z=18 10.0% vs 0.0%; rebate of z=17 6.9% vs 0.0%; compensation of z=17 6.5% vs 0.0%; reimbursement of z=17 5.9% vs 0.0%; restitution of z=16 5.3% vs 0.0%; refund of z=16 5.8% vs 0.0%; a rebate of z=15 5.6% vs 0.0%; request a z=15 5.1% vs 0.2%; ask for a z=15 4.9% vs 0.1%; my money back z=15 4.6% vs 0.1%; a compensation of z=15 6.2% vs 0.0%

2–4-word phrases, get_weather: the weather z=56 35.8% vs 0.4%; what is the z=36 17.3% vs 1.4%; what's the z=35 12.3% vs 0.3%; is the z=34 18.8% vs 2.3%; it be z=33 12.9% vs 0.2%; will it z=33 16.8% vs 0.1%; what is z=33 17.7% vs 2.2%; is it z=31 12.8% vs 1.1%; is the weather z=30 9.5% vs 0.2%; will the z=28 9.9% vs 0.1%; what is the weather z=28 8.5% vs 0.1%; from now z=27 7.6% vs 0.2%; forecast for z=26 20.5% vs 0.0%; will it be z=26 12.8% vs 0.0%; what will z=26 9.3% vs 0.1%; weather forecast z=25 12.8% vs 0.0%; to be z=25 7.3% vs 0.5%; going to z=24 6.9% vs 0.4%; what's the weather z=24 10.1% vs 0.0%; weather in z=24 5.8% vs 0.1%

2–4-word phrases, iot_cleaning: turn on z=16 16.1% vs 0.2%; turn on the z=15 11.0% vs 0.1%; start the z=13 8.5% vs 0.1%; on the z=11 13.6% vs 0.9%; activate the z=9 5.9% vs 0.0%; my robot z=8 3.4% vs 0.0%; the living room z=8 4.2% vs 0.1%; please start z=8 3.4% vs 0.0%; to start z=8 3.4% vs 0.0%; the living z=8 4.2% vs 0.1%; living room z=8 4.2% vs 0.1%; you turn on z=7 2.5% vs 0.0%; you turn on the z=7 2.5% vs 0.0%; clean the z=7 5.1% vs 0.0%; you turn z=6 2.5% vs 0.0%; please turn on z=6 2.5% vs 0.1%; to turn on z=6 1.7% vs 0.0%; to turn on the z=6 1.7% vs 0.0%; clean my z=6 1.7% vs 0.0%; can you turn on z=6 1.7% vs 0.0%

2–4-word phrases, iot_coffee: make me z=20 16.7% vs 0.1%; please make z=14 8.3% vs 0.0%; make me a z=14 7.6% vs 0.1%; a cup z=13 10.4% vs 0.0%; cup of z=12 13.9% vs 0.0%; of coffee z=11 16.0% vs 0.0%; cup of coffee z=11 13.2% vs 0.0%; my coffee z=11 6.9% vs 0.0%; a cup of z=10 10.4% vs 0.0%; me some z=10 5.6% vs 0.1%; me a z=9 9.7% vs 0.9%; can you make z=9 3.5% vs 0.0%; the coffee z=9 13.2% vs 0.0%; please make me z=9 3.5% vs 0.0%; you make z=8 3.5% vs 0.1%; a cup of coffee z=8 10.4% vs 0.0%; coffee at z=8 6.2% vs 0.0%; make a z=7 6.9% vs 0.6%; like a z=7 3.5% vs 0.1%; to make z=7 5.6% vs 0.4%

2–4-word phrases, iot_hue_lightchange: change the z=28 28.2% vs 0.1%; the lights z=24 22.4% vs 0.2%; the light z=20 14.1% vs 0.1%; lights to z=18 15.3% vs 0.0%; make the z=17 10.6% vs 0.0%; the house z=17 10.6% vs 0.0%; lights in z=12 6.5% vs 0.1%; the lights in z=12 5.9% vs 0.1%; in the house z=12 5.9% vs 0.0%; please change z=12 5.3% vs 0.0%; color of z=12 10.0% vs 0.0%; the living room z=12 5.3% vs 0.1%; the living z=11 5.3% vs 0.1%; room lights z=11 4.7% vs 0.0%; living room z=11 5.3% vs 0.1%; light to z=11 5.9% vs 0.0%; make the lights z=11 5.3% vs 0.0%; to red z=10 7.6% vs 0.0%; lights in my z=10 3.5% vs 0.0%; the lights to z=10 3.5% vs 0.0%

2–4-word phrases, iot_hue_lightdim: the lights z=29 49.0% vs 0.2%; lights in z=17 16.3% vs 0.1%; lower the z=16 18.4% vs 0.0%; the lights in z=15 13.3% vs 0.1%; lights in the z=14 11.2% vs 0.0%; the lights in the z=14 11.2% vs 0.0%; down the z=13 9.2% vs 0.0%; living room z=12 9.2% vs 0.1%; turn down z=12 8.2% vs 0.0%; turn down the z=11 8.2% vs 0.0%; the living room z=11 7.1% vs 0.1%; the living z=10 7.1% vs 0.1%; in the living room z=9 5.1% vs 0.0%; in the living z=9 5.1% vs 0.0%; the light z=9 7.1% vs 0.1%; turn the lights z=9 5.1% vs 0.0%; the brightness z=8 4.1% vs 0.0%; the lights to z=8 4.1% vs 0.0%; please lower z=8 4.1% vs 0.0%; lights to z=8 5.1% vs 0.1%

2–4-word phrases, iot_hue_lightoff: turn off z=34 45.4% vs 0.1%; off the z=27 26.4% vs 0.1%; the lights z=24 22.4% vs 0.2%; turn off the z=23 21.3% vs 0.1%; please turn z=21 15.5% vs 0.1%; lights in z=18 11.5% vs 0.1%; please turn off z=17 12.6% vs 0.0%; the light z=17 10.3% vs 0.1%; turn the z=16 9.8% vs 0.1%; living room z=15 8.0% vs 0.1%; the living z=14 7.5% vs 0.1%; the lights in z=14 6.9% vs 0.1%; the living room z=13 6.3% vs 0.1%; you turn z=13 5.7% vs 0.0%; room lights z=13 5.7% vs 0.0%; the kitchen z=12 5.7% vs 0.0%; turn the lights z=12 5.2% vs 0.0%; lights in the z=12 5.2% vs 0.0%; please turn off the z=12 5.7% vs 0.0%; switch off z=12 5.2% vs 0.0%

2–4-word phrases, iot_hue_lighton: lights on z=11 22.2% vs 0.0%; turn on z=10 33.3% vs 0.3%; turn on the z=8 18.5% vs 0.1%; the lights on z=8 11.1% vs 0.0%; the lights z=7 22.2% vs 0.3%; it's too z=7 11.1% vs 0.0%; turn the lights z=7 11.1% vs 0.0%; please turn on z=7 11.1% vs 0.1%; on all z=6 7.4% vs 0.0%; please turn z=6 14.8% vs 0.2%; all lights z=6 7.4% vs 0.0%; in here z=6 7.4% vs 0.0%; turn the z=6 11.1% vs 0.1%; on the z=4 18.5% vs 0.9%; to turn on z=4 3.7% vs 0.0%; to turn on the z=4 3.7% vs 0.0%; in my house z=4 3.7% vs 0.0%; the blue z=4 3.7% vs 0.0%; the lamp z=4 3.7% vs 0.0%; please turn the lights z=4 3.7% vs 0.0%

2–4-word phrases, iot_hue_lightup: the lights z=21 27.2% vs 0.3%; the light z=17 16.5% vs 0.1%; turn up z=15 11.7% vs 0.0%; up the z=15 15.5% vs 0.3%; increase the z=14 10.7% vs 0.0%; turn up the z=14 10.7% vs 0.0%; lights in z=13 9.7% vs 0.1%; the room z=12 7.8% vs 0.0%; make the z=12 7.8% vs 0.1%; the lights in z=11 7.8% vs 0.1%; the brightness z=10 5.8% vs 0.0%; raise the z=9 4.9% vs 0.0%; this room z=9 4.9% vs 0.0%; living room z=9 5.8% vs 0.1%; lights please z=9 4.9% vs 0.0%; please turn up z=8 3.9% vs 0.0%; please turn up the z=8 3.9% vs 0.0%; a little z=8 3.9% vs 0.0%; in this room z=8 3.9% vs 0.0%; please increase z=7 2.9% vs 0.0%

2–4-word phrases, iot_wemo_off: turn off z=26 50.7% vs 0.2%; off the z=21 32.0% vs 0.1%; plug socket z=19 29.3% vs 0.0%; wemo plug z=18 25.3% vs 0.0%; turn off the z=18 22.7% vs 0.1%; the wemo z=16 18.7% vs 0.0%; wemo plug socket z=15 16.0% vs 0.0%; off my z=13 12.0% vs 0.0%; the wemo plug z=13 12.0% vs 0.0%; please turn off z=12 10.7% vs 0.0%; the plug z=12 10.7% vs 0.0%; switch off z=11 9.3% vs 0.0%; turn off my z=11 9.3% vs 0.0%; smart socket z=11 10.7% vs 0.0%; smart plug z=10 9.3% vs 0.0%; the smart z=10 10.7% vs 0.0%; please turn z=10 10.7% vs 0.2%; the wemo plug socket z=10 6.7% vs 0.0%; please turn off the z=10 6.7% vs 0.0%; switch off the z=9 8.0% vs 0.0%

2–4-word phrases, iot_wemo_on: turn on z=24 51.6% vs 0.2%; turn on the z=18 29.0% vs 0.1%; plug socket z=18 27.4% vs 0.0%; wemo plug z=17 24.2% vs 0.0%; wemo plug socket z=15 21.0% vs 0.0%; the wemo z=15 19.4% vs 0.0%; on the z=13 32.3% vs 0.9%; the wemo plug z=13 14.5% vs 0.0%; the wemo plug socket z=11 11.3% vs 0.0%; the plug z=11 9.7% vs 0.0%; please turn on z=10 9.7% vs 0.0%; smart plug z=9 6.5% vs 0.0%; smart socket z=9 6.5% vs 0.0%; turn on my z=9 6.5% vs 0.0%; please turn z=8 9.7% vs 0.2%; on the wemo z=8 12.9% vs 0.0%; on the wemo plug z=7 11.3% vs 0.0%; please turn on the z=7 4.8% vs 0.0%; switch on z=7 4.8% vs 0.0%; the smart z=7 4.8% vs 0.0%

2–4-word phrases, lists_createoradd: a new z=25 19.9% vs 0.3%; grocery list z=21 13.4% vs 0.1%; a list z=20 11.6% vs 0.1%; list of z=18 10.2% vs 0.1%; create a z=18 13.0% vs 0.4%; list for z=17 8.8% vs 0.0%; shopping list z=16 7.9% vs 0.1%; create a new z=15 7.9% vs 0.0%; my grocery list z=15 6.5% vs 0.0%; my grocery z=15 6.5% vs 0.0%; to my z=14 16.7% vs 1.8%; the list z=14 7.4% vs 0.2%; to do list z=13 5.6% vs 0.1%; do list z=13 5.6% vs 0.1%; make a z=12 9.3% vs 0.6%; a list of z=12 4.6% vs 0.1%; start a z=12 4.2% vs 0.0%; my shopping list z=12 4.2% vs 0.0%; my shopping z=12 4.2% vs 0.0%; my to do list z=12 4.2% vs 0.0%

2–4-word phrases, lists_query: the list z=24 15.1% vs 0.1%; my list z=24 14.4% vs 0.1%; on my z=23 19.6% vs 0.8%; do list z=17 7.0% vs 0.0%; shopping list z=16 7.0% vs 0.1%; to do list z=16 6.6% vs 0.0%; on the list z=15 7.0% vs 0.0%; list for z=15 5.5% vs 0.1%; is on z=15 5.5% vs 0.1%; what is on z=13 4.8% vs 0.0%; my to do z=13 4.4% vs 0.0%; my to do list z=13 4.4% vs 0.0%; my to z=13 4.4% vs 0.1%; grocery list z=13 4.8% vs 0.1%; my shopping list z=13 4.1% vs 0.0%; my shopping z=13 4.1% vs 0.0%; what's on z=13 4.1% vs 0.0%; do i have z=12 9.6% vs 0.9%; on the z=12 9.2% vs 0.8%; on my list z=12 6.3% vs 0.0%

2–4-word phrases, lists_remove: the list z=25 17.0% vs 0.1%; from my z=21 13.4% vs 0.2%; please remove z=18 8.9% vs 0.0%; shopping list z=18 8.5% vs 0.1%; remove the z=17 8.5% vs 0.1%; delete the z=17 8.0% vs 0.1%; grocery list z=17 8.0% vs 0.1%; item from z=14 5.4% vs 0.0%; my list z=14 6.2% vs 0.1%; my grocery list z=13 4.9% vs 0.0%; my grocery z=13 4.9% vs 0.0%; please delete z=12 4.0% vs 0.0%; from the z=12 8.0% vs 0.6%; to do list z=12 4.0% vs 0.1%; do list z=12 4.0% vs 0.1%; remove my z=11 4.0% vs 0.1%; my shopping list z=11 3.6% vs 0.0%; my shopping z=11 3.6% vs 0.0%; my to z=11 3.6% vs 0.1%; rid of z=11 3.1% vs 0.0%

2–4-word phrases, music_likeness: this song z=25 27.3% vs 0.2%; i like z=20 16.7% vs 0.1%; i love z=15 9.1% vs 0.0%; that song z=13 7.6% vs 0.0%; song to z=12 8.3% vs 0.2%; save this z=11 9.8% vs 0.0%; this song to z=11 6.1% vs 0.1%; my favorite z=11 6.1% vs 0.1%; song to my z=11 6.1% vs 0.1%; this song to my z=10 4.5% vs 0.0%; my favorites z=9 3.8% vs 0.0%; song is z=9 3.8% vs 0.0%; save the z=8 3.0% vs 0.0%; song in my z=8 3.0% vs 0.0%; this is z=8 6.1% vs 0.3%; rock music z=8 3.0% vs 0.0%; i really z=8 3.0% vs 0.0%; this is a z=8 3.0% vs 0.0%; like this z=8 3.0% vs 0.0%; song in z=8 3.0% vs 0.0%

2–4-word phrases, music_query: this song z=24 19.9% vs 0.2%; the song z=21 14.9% vs 0.2%; the name z=18 10.0% vs 0.1%; name of z=17 10.0% vs 0.0%; who is z=16 8.5% vs 0.1%; the name of z=16 9.5% vs 0.0%; of this z=16 8.5% vs 0.1%; is this z=16 8.0% vs 0.1%; song is z=13 6.0% vs 0.0%; who is the z=13 5.5% vs 0.0%; the artist z=12 5.5% vs 0.1%; name of the z=12 6.0% vs 0.0%; is playing z=12 5.0% vs 0.1%; that song z=11 4.0% vs 0.0%; the name of the z=11 5.5% vs 0.0%; currently playing z=11 4.0% vs 0.0%; the singer z=11 4.0% vs 0.0%; is the z=11 16.4% vs 2.8%; of the z=11 10.9% vs 1.3%; of this song z=10 5.5% vs 0.0%

2–4-word phrases, music_settings: repeat the z=11 9.7% vs 0.0%; last song z=9 6.5% vs 0.0%; the last song z=9 6.5% vs 0.0%; repeat this z=9 6.5% vs 0.0%; repeat the last z=8 4.8% vs 0.0%; the music z=7 6.5% vs 0.1%; this song z=7 9.7% vs 0.3%; to the next z=7 4.8% vs 0.0%; music player z=7 4.8% vs 0.0%; this playlist z=6 3.2% vs 0.0%; the next song z=6 3.2% vs 0.0%; on repeat z=6 3.2% vs 0.0%; please repeat z=6 3.2% vs 0.0%; this music z=6 3.2% vs 0.0%; current song z=6 3.2% vs 0.0%; next song z=6 3.2% vs 0.0%; please put z=5 3.2% vs 0.1%; the last z=4 6.5% vs 0.5%; skip to the next z=4 1.6% vs 0.0%; song one z=4 1.6% vs 0.0%

2–4-word phrases, news_query: the latest z=33 15.2% vs 0.1%; news on z=20 5.3% vs 0.1%; me the z=19 10.2% vs 1.7%; the news z=18 9.8% vs 0.0%; what is z=17 11.3% vs 2.6%; the world z=16 3.4% vs 0.1%; give me z=15 5.4% vs 0.6%; going on z=15 3.4% vs 0.1%; tell me z=15 8.2% vs 1.7%; in the world z=15 2.9% vs 0.1%; me the latest z=15 4.1% vs 0.0%; on the z=15 5.9% vs 0.8%; happening in z=14 2.6% vs 0.0%; are the z=14 4.0% vs 0.4%; what are the z=14 3.7% vs 0.3%; read me z=13 2.3% vs 0.0%; what are z=13 4.1% vs 0.5%; news from z=13 6.9% vs 0.0%; on in z=13 2.0% vs 0.0%; in the z=13 6.1% vs 1.4%

2–4-word phrases, newsletter_subscription: to the z=28 24.7% vs 1.7%; sign up z=24 13.8% vs 0.6%; to receive z=21 8.3% vs 0.1%; up to z=20 15.8% vs 0.0%; cancel my z=20 8.0% vs 0.1%; to cancel z=20 8.8% vs 0.3%; cancel the z=19 7.3% vs 0.1%; to your z=18 19.0% vs 0.0%; your company z=16 7.0% vs 0.4%; to cancel my z=15 4.5% vs 0.0%; to cancel the z=15 4.3% vs 0.0%; to sign z=14 8.9% vs 0.0%; to sign up z=14 8.9% vs 0.0%; i cancel z=12 2.7% vs 0.1%; ur company z=11 3.8% vs 0.2%; the status of my z=11 2.5% vs 0.1%; signing up z=11 2.3% vs 0.0%; help cancelling z=10 2.0% vs 0.0%; status of my z=10 2.8% vs 0.1%; up to the z=10 8.5% vs 0.0%

2–4-word phrases, payment_issue: inform of z=33 22.6% vs 0.4%; notify of z=33 22.2% vs 0.4%; to notify z=27 15.1% vs 0.2%; to notify of z=27 14.9% vs 0.2%; to inform z=27 14.9% vs 0.3%; to inform of z=27 14.8% vs 0.3%; to report z=27 14.3% vs 0.2%; of a z=26 15.7% vs 0.5%; problems with z=23 11.2% vs 0.3%; of an z=23 10.6% vs 0.2%; issues with z=22 10.1% vs 0.1%; errors with z=22 9.4% vs 0.1%; error with z=21 9.0% vs 0.1%; an error z=21 8.9% vs 0.2%; issue with z=21 8.7% vs 0.2%; a trouble z=20 8.6% vs 0.1%; an error with z=20 7.9% vs 0.1%; a problem with z=19 7.8% vs 0.2%; an issue with z=18 7.1% vs 0.2%; a problem z=18 8.3% vs 0.3%

2–4-word phrases, place_order: to order z=22 8.8% vs 0.2%; to purchase z=22 8.8% vs 0.1%; some of z=20 14.5% vs 0.0%; to buy z=20 8.2% vs 0.1%; an article z=17 5.2% vs 0.1%; several articles z=16 4.9% vs 0.0%; several products z=16 4.9% vs 0.0%; some articles z=16 4.5% vs 0.1%; some items z=15 4.4% vs 0.1%; several items z=14 3.5% vs 0.1%; order some z=13 3.9% vs 0.0%; some products z=13 3.3% vs 0.1%; order a z=13 3.2% vs 0.0%; a product z=13 5.5% vs 0.4%; an item z=13 3.8% vs 0.2%; of your z=13 26.6% vs 0.0%; to shop z=12 8.8% vs 0.0%; to do to z=12 11.9% vs 2.4%; do to z=12 11.9% vs 2.5%; to order some z=12 2.8% vs 0.0%

2–4-word phrases, play_audiobook: where i z=14 6.4% vs 0.0%; i left z=11 5.7% vs 0.0%; harry potter z=10 4.5% vs 0.0%; play the z=10 9.6% vs 0.9%; left off z=10 7.0% vs 0.0%; you play my z=8 2.5% vs 0.0%; can you play my z=8 2.5% vs 0.0%; the two z=8 2.5% vs 0.0%; book of z=8 2.5% vs 0.0%; last played z=8 2.5% vs 0.0%; can you play z=8 3.8% vs 0.2%; you play z=8 3.8% vs 0.2%; i left off z=8 5.7% vs 0.0%; resume the z=8 5.7% vs 0.0%; where i left z=8 5.7% vs 0.0%; where i left off z=8 5.7% vs 0.0%; from where z=7 1.9% vs 0.0%; playing my z=7 1.9% vs 0.0%; the god z=7 1.9% vs 0.0%; the audio z=7 1.9% vs 0.0%

2–4-word phrases, play_game: with me z=24 18.4% vs 0.1%; to play z=19 12.6% vs 0.2%; a game z=19 11.5% vs 0.0%; let's play z=15 9.8% vs 0.0%; want to play z=15 6.9% vs 0.0%; game of z=13 7.5% vs 0.0%; play a z=13 9.8% vs 0.5%; like to play z=11 4.0% vs 0.0%; i want to play z=10 3.4% vs 0.0%; play with z=10 4.6% vs 0.0%; with you z=10 4.0% vs 0.1%; to play a z=9 2.9% vs 0.0%; please open z=8 2.3% vs 0.0%; the game z=8 2.9% vs 0.1%; you play z=8 3.4% vs 0.2%; me the game z=7 1.7% vs 0.0%; play for me the z=7 1.7% vs 0.0%; game with z=7 3.4% vs 0.0%; play for me z=7 1.7% vs 0.0%; for me the z=7 1.7% vs 0.0%

2–4-word phrases, play_music: play the z=44 11.1% vs 0.3%; from the z=36 7.0% vs 0.3%; to hear z=33 6.0% vs 0.2%; play a z=31 8.6% vs 0.1%; play me z=27 5.4% vs 0.1%; want to hear z=27 4.7% vs 0.1%; i want to hear z=27 4.6% vs 0.1%; please play z=26 4.1% vs 0.1%; listen to z=23 2.9% vs 0.1%; i want to z=23 6.1% vs 2.2%; want to z=23 6.3% vs 2.3%; play my z=23 3.9% vs 0.0%; a song z=21 2.6% vs 0.1%; you play z=21 2.8% vs 0.0%; to listen z=20 2.2% vs 0.1%; and play z=20 2.3% vs 0.1%; to play z=20 2.1% vs 0.1%; can you play z=19 2.5% vs 0.0%; to listen to z=19 2.0% vs 0.1%; the top z=19 1.9% vs 0.1%

2–4-word phrases, play_podcasts: the next z=25 16.4% vs 0.3%; play the z=19 14.5% vs 0.8%; please play z=16 8.2% vs 0.3%; episode of z=14 11.9% vs 0.0%; please play the z=14 4.5% vs 0.0%; play next z=13 5.6% vs 0.0%; next episode z=12 20.1% vs 0.0%; to the next z=12 6.7% vs 0.0%; play the next z=11 5.9% vs 0.0%; next episode of z=11 9.3% vs 0.0%; start the z=11 3.0% vs 0.1%; go to z=10 4.5% vs 0.3%; listen to z=9 3.7% vs 0.2%; the next episode z=9 13.0% vs 0.0%; to the z=9 8.9% vs 2.1%; to next z=9 3.3% vs 0.0%; the next episode of z=8 6.3% vs 0.0%; go to the z=8 2.2% vs 0.1%; episode of the z=8 2.2% vs 0.0%; skip to z=8 2.2% vs 0.0%

2–4-word phrases, play_radio: turn on z=24 10.8% vs 0.2%; hundred and z=21 7.6% vs 0.1%; on the z=20 11.6% vs 0.8%; the radio z=18 15.9% vs 0.0%; play the z=18 9.8% vs 0.8%; to listen z=15 5.0% vs 0.2%; turn on the z=15 4.3% vs 0.1%; listen to z=15 5.0% vs 0.2%; nine hundred and z=14 3.8% vs 0.0%; nine hundred z=14 3.8% vs 0.0%; and play z=14 4.3% vs 0.1%; to listen to z=14 4.0% vs 0.1%; on the radio z=13 7.8% vs 0.0%; please play z=12 4.0% vs 0.3%; one hundred z=12 2.5% vs 0.0%; put on z=12 2.5% vs 0.1%; one hundred and z=11 2.3% vs 0.0%; want to listen z=11 2.8% vs 0.1%; i want to listen z=11 2.8% vs 0.1%; please turn on z=11 2.3% vs 0.0%

2–4-word phrases, qa_currency: how much z=22 13.7% vs 0.1%; what is z=21 29.2% vs 2.6%; what is the z=20 23.9% vs 1.9%; is the z=19 27.1% vs 2.8%; how many z=19 12.0% vs 0.3%; how much is z=18 9.5% vs 0.0%; much is z=18 9.5% vs 0.0%; u s z=17 16.9% vs 0.0%; rate of z=15 9.9% vs 0.0%; the us z=15 6.3% vs 0.0%; exchange rate z=13 37.0% vs 0.0%; rate for z=13 7.0% vs 0.0%; the exchange z=12 26.1% vs 0.0%; the exchange rate z=12 22.5% vs 0.0%; and the z=11 4.9% vs 0.2%; is one z=11 3.2% vs 0.0%; between the z=11 4.2% vs 0.0%; what is the current z=10 3.5% vs 0.1%; is the current z=10 3.5% vs 0.1%; exchange rate for z=10 6.3% vs 0.0%

2–4-word phrases, qa_definition: what does z=26 14.0% vs 0.1%; what is z=21 19.2% vs 2.6%; what is a z=17 6.1% vs 0.0%; is a z=16 6.1% vs 0.2%; meaning of z=14 4.7% vs 0.0%; the word z=13 7.9% vs 0.0%; the meaning z=13 3.5% vs 0.0%; the meaning of z=12 3.2% vs 0.0%; what a z=12 3.8% vs 0.0%; look like z=12 2.6% vs 0.0%; look up z=11 3.2% vs 0.1%; me what z=11 4.1% vs 0.3%; give me the z=11 3.8% vs 0.3%; tell me what z=11 3.5% vs 0.2%; what is the z=10 7.9% vs 1.9%; tell me z=10 7.3% vs 1.8%; definition of z=10 14.0% vs 0.0%; does a z=9 1.7% vs 0.0%; give me z=9 4.4% vs 0.6%; the definition z=9 12.5% vs 0.0%

2–4-word phrases, qa_factoid: is the z=31 18.2% vs 2.7%; how many z=28 8.2% vs 0.3%; what is z=26 14.4% vs 2.5%; where is z=25 7.0% vs 0.3%; what is the z=23 11.2% vs 1.9%; the world z=20 4.0% vs 0.1%; who is z=19 3.6% vs 0.1%; in the world z=18 3.5% vs 0.0%; tell me z=18 7.8% vs 1.7%; tell me about z=16 3.4% vs 0.2%; in the z=14 5.7% vs 1.4%; me about z=13 3.4% vs 0.5%; united states z=13 1.7% vs 0.0%; how old z=12 4.3% vs 0.0%; when was z=12 1.6% vs 0.0%; are the z=12 2.8% vs 0.4%; old is z=12 3.9% vs 0.0%; how old is z=12 3.9% vs 0.0%; was the z=12 1.6% vs 0.1%; the highest z=11 1.3% vs 0.0%

2–4-word phrases, qa_maths: what is z=18 43.2% vs 2.6%; the answer z=14 10.8% vs 0.1%; answer to z=13 9.0% vs 0.0%; the answer to z=13 9.0% vs 0.0%; one hundred z=11 6.3% vs 0.0%; is the answer z=10 8.1% vs 0.0%; what is the z=10 19.8% vs 2.0%; what is the answer z=10 7.2% vs 0.0%; hundred and z=10 7.2% vs 0.2%; is the z=9 20.7% vs 2.9%; one hundred and z=9 4.5% vs 0.0%; is the answer to z=9 7.2% vs 0.0%; to two z=9 4.5% vs 0.0%; is twelve z=8 3.6% vs 0.0%; is two z=8 3.6% vs 0.0%; is one z=8 3.6% vs 0.0%; divided by z=8 8.1% vs 0.0%; five and z=7 2.7% vs 0.0%; is twenty z=7 2.7% vs 0.0%; hundred and twenty z=7 2.7% vs 0.0%

2–4-word phrases, qa_stock: price of z=19 18.8% vs 0.0%; what is z=14 20.3% vs 2.7%; price for z=13 11.2% vs 0.0%; how much z=12 6.6% vs 0.2%; what is the z=12 14.7% vs 1.9%; is the z=12 17.3% vs 2.8%; the price z=10 4.1% vs 0.0%; stock price z=10 25.9% vs 0.0%; the price of z=9 3.6% vs 0.0%; stock price for z=8 10.2% vs 0.0%; much does z=8 2.5% vs 0.0%; how much does z=8 2.5% vs 0.0%; is the price z=8 2.5% vs 0.0%; what is the price z=8 2.5% vs 0.0%; is the current z=8 3.0% vs 0.1%; out the z=8 2.0% vs 0.0%; of google z=8 2.0% vs 0.0%; find out z=8 2.0% vs 0.0%; much is z=8 2.5% vs 0.1%; how much is z=8 2.5% vs 0.1%

2–4-word phrases, rate_book: the current z=36 13.5% vs 0.5%; out of z=26 34.2% vs 0.0%; of 6 z=25 50.8% vs 0.0%; the book z=21 4.4% vs 0.1%; this album z=19 3.7% vs 0.1%; and a z=19 3.7% vs 0.1%; i give z=18 7.0% vs 0.0%; give the z=18 10.7% vs 0.0%; to give z=17 2.9% vs 0.1%; to this z=15 2.4% vs 0.1%; i would z=14 4.2% vs 0.7%; value of z=14 2.2% vs 0.0%; rating of z=13 7.8% vs 0.0%; rate this z=12 14.2% vs 0.0%; a best z=12 3.5% vs 0.0%; a two z=11 1.8% vs 0.0%; this current z=11 4.1% vs 0.0%; book the z=11 1.5% vs 0.1%; one of z=10 1.2% vs 0.0%; the book the z=10 1.4% vs 0.0%

2–4-word phrases, recommendation_events: this weekend z=22 13.0% vs 0.1%; going on z=22 12.6% vs 0.1%; there any z=16 8.9% vs 0.3%; happening in z=16 6.9% vs 0.0%; events in z=16 6.9% vs 0.0%; what events z=15 6.5% vs 0.0%; events are z=14 7.7% vs 0.0%; on in z=14 5.3% vs 0.0%; going on in z=14 5.3% vs 0.0%; what events are z=13 6.1% vs 0.0%; any events z=13 4.9% vs 0.0%; my area z=13 4.9% vs 0.0%; in my z=13 8.1% vs 0.5%; in my area z=12 4.5% vs 0.0%; is happening z=12 4.0% vs 0.0%; what is happening z=12 3.6% vs 0.0%; is going on z=12 3.6% vs 0.0%; are there any z=12 4.9% vs 0.2%; is there z=11 6.9% vs 0.5%; is there any z=11 4.0% vs 0.1%

2–4-word phrases, recommendation_locations: where is z=19 14.7% vs 0.3%; where is the z=17 9.9% vs 0.1%; a good z=13 6.3% vs 0.1%; near me z=13 5.8% vs 0.1%; is the nearest z=12 5.2% vs 0.1%; where is the nearest z=12 4.7% vs 0.0%; the best z=11 6.3% vs 0.2%; the nearest z=11 6.3% vs 0.3%; local area z=9 2.6% vs 0.0%; restaurants in z=9 3.1% vs 0.0%; near my z=9 2.6% vs 0.0%; this area z=9 2.6% vs 0.0%; place to z=9 4.2% vs 0.0%; where is the closest z=9 2.6% vs 0.0%; is a z=9 4.7% vs 0.3%; the closest z=9 4.7% vs 0.3%; in this z=9 3.1% vs 0.1%; in this area z=8 2.6% vs 0.0%; is the closest z=8 2.6% vs 0.1%; me where z=8 2.1% vs 0.0%

2–4-word phrases, recommendation_movies: what movies z=15 15.9% vs 0.1%; what movies are z=14 14.6% vs 0.1%; movies are z=14 17.1% vs 0.2%; what movies are playing z=12 9.8% vs 0.0%; are playing z=12 12.2% vs 0.2%; movies are playing z=11 9.8% vs 0.1%; playing in z=11 8.5% vs 0.1%; near me z=9 7.3% vs 0.1%; a good z=9 7.3% vs 0.1%; what movie z=9 6.1% vs 0.1%; what new z=7 3.7% vs 0.0%; recommend a z=7 3.7% vs 0.0%; i watch z=7 4.9% vs 0.1%; should i z=7 6.1% vs 0.1%; movies playing z=7 4.9% vs 0.1%; playing near z=7 3.7% vs 0.0%; playing near me z=7 3.7% vs 0.0%; in me area z=7 3.7% vs 0.0%; me area z=7 3.7% vs 0.0%; movies playing in z=7 3.7% vs 0.0%

2–4-word phrases, recover_password: of my z=42 36.1% vs 1.0%; my user z=40 40.5% vs 0.2%; my account z=24 15.9% vs 0.0%; my profile z=23 15.6% vs 0.0%; user account z=22 14.2% vs 0.0%; user profile z=22 10.2% vs 0.1%; my user account z=21 14.0% vs 0.0%; my user profile z=20 9.7% vs 0.0%; information about z=15 7.9% vs 0.6%; of my user z=12 18.8% vs 0.0%; information about my z=12 2.7% vs 0.1%; about my z=11 3.4% vs 0.2%; lost my z=11 2.2% vs 0.0%; question about z=10 2.7% vs 0.1%; information about a z=10 2.2% vs 0.1%; reset my z=10 8.5% vs 0.0%; reset the z=10 7.4% vs 0.0%; find information about z=9 2.2% vs 0.1%; reset it z=9 2.7% vs 0.0%; find information z=9 2.2% vs 0.1%

2–4-word phrases, registration_problems: sign up z=40 35.7% vs 0.2%; inform of z=36 25.8% vs 0.3%; notify of z=35 23.9% vs 0.3%; with a z=33 22.7% vs 0.5%; to inform z=29 16.7% vs 0.2%; to inform of z=29 16.6% vs 0.2%; to report z=28 15.3% vs 0.2%; to notify z=28 15.3% vs 0.2%; to notify of z=27 15.1% vs 0.2%; of a z=27 16.6% vs 0.5%; of an z=25 12.2% vs 0.2%; issue with z=23 10.9% vs 0.1%; an issue z=22 10.4% vs 0.2%; an issue with z=22 9.8% vs 0.1%; error with z=20 8.2% vs 0.1%; a problem with z=20 8.6% vs 0.2%; an error with z=20 8.0% vs 0.1%; an error z=20 8.2% vs 0.2%; inform of a z=19 7.6% vs 0.1%; a problem z=19 9.0% vs 0.3%

2–4-word phrases, review: a product z=30 19.3% vs 0.2%; about a z=29 17.5% vs 0.2%; for a z=27 19.6% vs 0.8%; to submit z=25 12.7% vs 0.2%; about your z=24 13.1% vs 0.1%; to send z=23 17.2% vs 0.0%; for your z=23 16.3% vs 0.0%; your company z=22 11.3% vs 0.3%; is there z=17 8.3% vs 0.4%; ur company z=16 6.3% vs 0.2%; write a z=15 7.3% vs 0.0%; to leave z=15 26.0% vs 0.0%; for a product z=15 9.9% vs 0.0%; there a z=15 4.6% vs 0.1%; is there a z=15 4.6% vs 0.1%; submit my z=14 4.1% vs 0.1%; your services z=14 8.8% vs 0.0%; do you have z=14 4.0% vs 0.1%; to write z=13 8.5% vs 0.0%; to submit my z=13 3.4% vs 0.0%

2–4-word phrases, search_creative_work: find the z=47 17.3% vs 0.5%; find a z=38 11.7% vs 0.3%; me the z=32 11.1% vs 1.4%; find me z=26 5.4% vs 0.2%; where can z=25 6.1% vs 0.7%; where can i z=25 6.1% vs 0.7%; show me z=24 6.6% vs 0.8%; show me the z=24 4.9% vs 0.3%; for the z=24 6.4% vs 0.9%; find me the z=22 4.3% vs 0.1%; show the z=20 3.8% vs 0.1%; search for z=20 3.1% vs 0.1%; looking for z=19 3.2% vs 0.2%; of the z=18 5.1% vs 1.2%; can i find z=17 2.4% vs 0.1%; i find z=17 2.4% vs 0.1%; i'm looking z=17 2.3% vs 0.1%; where can i find z=17 2.4% vs 0.0%; look up z=17 2.2% vs 0.1%; you find z=17 2.3% vs 0.0%

2–4-word phrases, search_screening_event: at the z=34 13.0% vs 0.5%; the movie z=33 19.2% vs 0.1%; what time z=26 9.6% vs 0.7%; what time is z=26 7.5% vs 0.2%; in the z=26 11.3% vs 1.2%; time is z=26 7.5% vs 0.2%; schedule for z=25 10.2% vs 0.1%; find the z=24 8.8% vs 0.8%; the nearest z=23 7.6% vs 0.1%; the closest z=22 7.6% vs 0.0%; is the z=21 13.1% vs 2.6%; movies are z=18 5.9% vs 0.0%; are playing z=18 4.6% vs 0.0%; close by z=17 3.4% vs 0.1%; the schedule z=17 5.1% vs 0.0%; to watch z=17 3.1% vs 0.1%; playing at z=17 11.6% vs 0.0%; a movie z=17 3.3% vs 0.0%; when is z=16 3.4% vs 0.2%; times for z=16 4.4% vs 0.0%

2–4-word phrases, set_up_shipping_address: shipping address z=50 48.2% vs 0.4%; delivery address z=48 45.1% vs 0.4%; set up z=24 13.4% vs 0.1%; to submit z=24 11.0% vs 0.2%; the new z=21 8.6% vs 0.1%; to set z=21 18.3% vs 0.0%; up my z=20 7.8% vs 0.1%; my new z=20 8.4% vs 0.0%; up a z=19 7.3% vs 0.1%; a new z=17 6.8% vs 0.3%; my shipping address z=17 5.9% vs 0.2%; my shipping z=17 5.9% vs 0.2%; set my z=17 5.5% vs 0.0%; up the z=17 6.0% vs 0.2%; a different z=16 6.5% vs 0.0%; submit a z=15 5.8% vs 0.0%; submit my z=15 4.2% vs 0.1%; the delivery address z=15 4.4% vs 0.1%; the shipping address z=14 4.7% vs 0.2%; set the z=14 4.0% vs 0.1%

2–4-word phrases, social_post: a complaint z=25 11.1% vs 0.2%; send a z=22 8.3% vs 0.1%; customer service z=17 8.1% vs 0.5%; on twitter z=15 4.5% vs 0.0%; on facebook z=15 4.3% vs 0.0%; that i z=12 3.0% vs 0.1%; my facebook z=11 2.3% vs 0.0%; and tell z=11 2.3% vs 0.0%; that the z=11 2.0% vs 0.0%; message to z=10 1.8% vs 0.0%; write a z=10 2.3% vs 0.1%; status on z=10 2.3% vs 0.0%; post on z=9 1.5% vs 0.0%; that i am z=9 1.8% vs 0.0%; tell them z=8 2.0% vs 0.0%; a message z=8 1.3% vs 0.0%; for me the z=8 1.3% vs 0.0%; my twitter z=8 1.3% vs 0.0%; happy birthday z=8 1.3% vs 0.0%; and tell them z=8 1.8% vs 0.0%

2–4-word phrases, social_query: my facebook z=16 11.0% vs 0.0%; on facebook z=15 9.7% vs 0.0%; social media z=15 15.2% vs 0.0%; on twitter z=14 8.3% vs 0.0%; me what's z=11 4.8% vs 0.0%; tell me what's z=11 4.8% vs 0.0%; new on z=10 5.5% vs 0.1%; any new z=9 5.5% vs 0.2%; what's new z=9 3.4% vs 0.0%; post on z=9 3.4% vs 0.0%; my twitter z=9 3.4% vs 0.0%; my social z=9 4.8% vs 0.0%; my social media z=9 4.8% vs 0.0%; on my facebook z=9 3.4% vs 0.0%; on my z=8 9.0% vs 0.8%; what are the latest z=8 2.8% vs 0.0%; are the latest z=8 2.8% vs 0.0%; there any z=8 5.5% vs 0.3%; in facebook z=8 2.8% vs 0.0%; happening on z=8 2.8% vs 0.0%

2–4-word phrases, switch_account: to the z=44 49.0% vs 1.5%; to change z=30 19.5% vs 0.3%; use the z=25 27.5% vs 0.0%; the standard z=22 14.7% vs 0.0%; to switch z=20 10.1% vs 0.0%; the standard account z=20 10.1% vs 0.0%; standard account z=19 11.7% vs 0.6%; pro account z=18 9.0% vs 0.4%; the platinum z=18 7.8% vs 0.0%; the pro z=17 9.4% vs 0.0%; gold account z=17 8.0% vs 0.3%; the premium z=16 8.8% vs 0.0%; switch to z=16 15.5% vs 0.0%; change to z=16 28.8% vs 0.0%; the pro account z=15 7.4% vs 0.0%; freemium account z=15 7.0% vs 0.3%; premium account z=15 7.2% vs 0.3%; platinum account z=15 7.4% vs 0.4%; the gold z=15 9.1% vs 0.0%; the platinum account z=15 5.3% vs 0.0%

2–4-word phrases, takeaway_order: order a z=17 11.0% vs 0.1%; to order z=14 11.0% vs 0.3%; order me z=13 12.3% vs 0.0%; order some z=13 6.8% vs 0.1%; to order some z=12 5.5% vs 0.0%; a pizza z=11 4.8% vs 0.0%; can i order z=10 4.1% vs 0.0%; want to order z=10 5.5% vs 0.0%; i want to order z=10 5.5% vs 0.0%; near me z=10 4.8% vs 0.1%; an order z=10 4.1% vs 0.1%; chinese food z=10 4.8% vs 0.0%; i order z=10 4.1% vs 0.1%; take out z=9 3.4% vs 0.0%; please order z=9 4.8% vs 0.0%; call the z=8 2.7% vs 0.0%; me some z=8 4.1% vs 0.1%; and order z=8 6.2% vs 0.0%; pizza hut z=7 2.1% vs 0.0%; order for z=7 4.1% vs 0.0%

2–4-word phrases, takeaway_query: my order z=19 12.7% vs 0.1%; the restaurant z=12 5.4% vs 0.0%; take out z=11 4.2% vs 0.0%; is my z=11 6.0% vs 0.2%; pizza hut z=10 3.6% vs 0.0%; find out z=9 3.0% vs 0.0%; olive garden z=9 3.0% vs 0.0%; pick up z=9 3.0% vs 0.0%; if the z=9 3.0% vs 0.0%; does this z=8 2.4% vs 0.0%; can i order z=8 2.4% vs 0.0%; status of my z=8 3.6% vs 0.2%; status of z=8 5.4% vs 0.5%; what is the status z=8 2.4% vs 0.0%; is the status z=8 2.4% vs 0.0%; the status of my z=7 3.0% vs 0.1%; the status z=7 4.2% vs 0.3%; can i get z=7 4.8% vs 0.4%; my food z=7 4.2% vs 0.0%; how long until z=7 1.8% vs 0.0%

2–4-word phrases, track_order: of order z=32 19.6% vs 0.2%; current status z=30 17.3% vs 0.2%; status of z=30 17.5% vs 0.3%; of purchase z=29 16.6% vs 0.2%; see the z=27 22.2% vs 1.2%; order ord z=24 11.7% vs 0.3%; purchase ord z=23 10.6% vs 0.2%; the status z=22 10.1% vs 0.2%; the status of z=22 10.0% vs 0.2%; to see the z=22 14.4% vs 0.8%; check the z=21 15.2% vs 1.1%; of the z=20 15.0% vs 1.1%; to see z=19 20.6% vs 2.6%; the order z=18 7.2% vs 0.0%; the purchase z=17 7.2% vs 0.0%; current status of z=16 5.4% vs 0.1%; to check the z=16 9.0% vs 0.6%; of order ord z=16 5.1% vs 0.1%; to check z=15 14.0% vs 1.8%; see the status of z=15 4.7% vs 0.1%

2–4-word phrases, track_refund: status of z=29 21.5% vs 0.3%; current status z=28 19.7% vs 0.2%; if there z=26 26.8% vs 0.1%; there are z=25 15.8% vs 0.1%; there is z=21 11.6% vs 0.2%; status of my z=20 11.0% vs 0.1%; the status of z=20 10.7% vs 0.2%; the status z=20 10.7% vs 0.2%; current status of z=19 9.2% vs 0.1%; see if z=19 10.8% vs 0.0%; news on z=19 8.8% vs 0.1%; are any z=18 15.6% vs 0.0%; the current status of z=18 8.8% vs 0.1%; the current status z=18 8.8% vs 0.1%; on my z=17 12.7% vs 0.7%; anything new z=16 7.9% vs 0.0%; if there are z=16 15.8% vs 0.0%; there are any z=16 15.6% vs 0.0%; updates on z=16 7.4% vs 0.0%; the status of my z=16 6.2% vs 0.1%

2–4-word phrases, transport_query: does the z=21 10.6% vs 0.1%; the train z=20 16.6% vs 0.0%; a train z=18 9.3% vs 0.2%; time does z=17 7.3% vs 0.0%; what time does z=17 7.3% vs 0.0%; train ticket z=17 8.3% vs 0.2%; train to z=16 7.3% vs 0.0%; time does the z=16 7.0% vs 0.0%; what time does the z=16 7.0% vs 0.0%; when is the z=14 5.3% vs 0.1%; a train ticket z=14 6.0% vs 0.2%; get to z=14 5.6% vs 0.0%; what time z=14 10.6% vs 1.0%; the next z=13 7.0% vs 0.3%; from here z=13 4.0% vs 0.0%; how do i get z=12 3.7% vs 0.0%; is the next z=12 3.7% vs 0.0%; do i get z=12 3.7% vs 0.0%; when is z=12 5.6% vs 0.3%; next train z=12 5.6% vs 0.0%

2–4-word phrases, transport_taxi: book an z=13 7.3% vs 0.0%; the airport z=12 8.0% vs 0.0%; for me z=12 13.9% vs 0.7%; me an z=12 6.6% vs 0.1%; to pick z=11 5.8% vs 0.0%; book me z=11 8.0% vs 0.2%; me a z=11 13.1% vs 0.8%; me up z=10 5.8% vs 0.1%; me to the z=9 3.6% vs 0.0%; need a z=9 8.8% vs 0.5%; get me z=9 6.6% vs 0.2%; book a z=9 14.6% vs 1.8%; take me z=8 3.6% vs 0.0%; i need a z=8 7.3% vs 0.4%; for me to z=8 2.9% vs 0.0%; ride to z=8 2.9% vs 0.0%; get me a z=8 3.6% vs 0.1%; book me a z=8 5.1% vs 0.2%; a ride z=8 3.6% vs 0.0%; to the z=8 13.9% vs 2.1%

2–4-word phrases, transport_ticket: train ticket z=31 62.6% vs 0.1%; a train z=30 50.0% vs 0.1%; a train ticket z=26 44.3% vs 0.0%; ticket to z=21 30.5% vs 0.0%; train ticket to z=20 25.9% vs 0.0%; a train ticket to z=18 20.1% vs 0.0%; new york z=17 14.9% vs 0.3%; to new z=17 12.1% vs 0.0%; me a z=17 21.3% vs 0.8%; book me z=16 12.6% vs 0.2%; to new york z=16 12.1% vs 0.0%; book me a z=16 11.5% vs 0.2%; me a train z=15 15.5% vs 0.0%; train to z=15 8.6% vs 0.0%; ticket for z=15 16.7% vs 0.0%; book a z=14 23.0% vs 1.7%; me a train ticket z=14 13.8% vs 0.0%; train ticket for z=13 11.5% vs 0.0%; a train ticket for z=11 9.8% vs 0.0%; can you book z=11 5.7% vs 0.0%

2–4-word phrases, transport_traffic: the traffic z=16 30.3% vs 0.0%; how is z=16 10.5% vs 0.1%; how is the z=14 7.9% vs 0.1%; is the z=13 23.0% vs 2.8%; there any z=12 8.6% vs 0.3%; right now z=11 7.9% vs 0.3%; to work z=11 4.6% vs 0.0%; are there any z=10 5.9% vs 0.2%; is there z=10 8.6% vs 0.5%; long will z=9 3.3% vs 0.0%; it take z=9 3.3% vs 0.0%; how long will z=9 3.3% vs 0.0%; are there z=9 6.6% vs 0.4%; is the traffic z=9 15.8% vs 0.0%; this morning z=9 3.3% vs 0.0%; what is z=8 14.5% vs 2.7%; is traffic z=8 9.2% vs 0.0%; traffic on z=8 7.9% vs 0.0%; what is the traffic z=8 5.9% vs 0.0%; traffic like z=7 4.6% vs 0.0%

2–4-word phrases, weather_query: the weather z=40 25.1% vs 1.3%; is the weather z=27 10.4% vs 0.3%; is it z=26 14.1% vs 1.3%; going to z=23 8.9% vs 0.5%; weather in z=23 7.2% vs 0.2%; is the z=23 16.3% vs 2.7%; what is the weather z=23 7.6% vs 0.3%; this week z=21 6.3% vs 0.2%; should i z=20 5.6% vs 0.1%; what is the z=20 12.0% vs 1.8%; what is z=20 13.8% vs 2.5%; the weather in z=19 5.1% vs 0.2%; it going to z=18 4.4% vs 0.2%; it going z=18 4.4% vs 0.2%; is it going z=17 4.3% vs 0.2%; is it going to z=17 4.3% vs 0.2%; weather like z=17 3.7% vs 0.1%; do i need z=16 5.3% vs 0.4%; going to be z=16 4.3% vs 0.2%; the weather like z=16 3.4% vs 0.1%

Strong phrase flags (in ≥5% of one class's rows and at ≥4× the rate in the others):

set_up_shipping_address: address 97.0% vs 1.7%
change_shipping_address: address 96.4% vs 1.8%
check_payment_methods: payment 96.1% vs 1.0%
contact_customer_service: customer 95.1% vs 0.5%
newsletter_subscription: newsletter 94.3% vs 0.0%
delete_account: account 91.5% vs 3.5%
complaint: a 90.8% vs 17.3%
iot_hue_lightoff: off 90.2% vs 0.4%
transport_ticket: train 89.1% vs 0.4%
iot_coffee: coffee 88.9% vs 0.0%
recover_password: my 88.7% vs 13.0%
create_account: account 86.7% vs 3.6%
lists_createoradd: list 86.6% vs 1.6%
datetime_convert: time 86.6% vs 2.1%
book_restaurant: a 86.0% vs 16.0%
iot_wemo_off: off 84.0% vs 0.6%
iot_wemo_on: on 82.3% vs 6.6%
transport_traffic: traffic 81.6% vs 0.0%
lists_remove: list 81.2% vs 1.6%
play_music: play 81.1% vs 1.3%
add_to_playlist: add 81.1% vs 0.9%
transport_ticket: ticket 78.7% vs 0.1%
general_confirm: please 77.9% vs 4.7%
general_confirm: you 77.9% vs 7.7%
alarm_set: alarm 77.8% vs 0.3%
email_sendemail: email 77.6% vs 0.5%
payment_issue: with 76.9% vs 4.3%
iot_cleaning: vacuum 72.9% vs 0.0%
takeaway_order: order 72.6% vs 2.3%
email_addcontact: email 71.6% vs 1.2%
edit_account: on 71.6% vs 5.9%
qa_stock: stock 71.1% vs 0.0%
general_joke: joke 70.7% vs 0.0%
switch_account: account 70.6% vs 3.9%
book_restaurant: for 69.9% vs 9.4%
alarm_remove: alarm 69.6% vs 0.5%
iot_hue_lightdim: lights 69.4% vs 0.5%
check_refund_policy: can 68.9% vs 9.5%
edit_account: account 68.1% vs 4.0%
play_game: play 67.2% vs 5.2%

Standard flags (owner's rule: a word or phrase in more than 2% of one class's rows, of whose rows at least 70% (and at least twice the base rate) belong to that class; the reviewer decides whether each is meaning or a shortcut):

all rows (50,000 rows): contact_customer_service: customer 95.1% of class, 76% of its 996 rows; newsletter_subscription: newsletter 94.3% of class, 100% of its 741 rows; iot_coffee: coffee 88.9% of class, 90% of its 143 rows; transport_traffic: traffic 81.6% of class, 95% of its 130 rows; play_music: play 81.1% of class, 78% of its 2724 rows; add_to_playlist: add 81.1% of class, 76% of its 1800 rows; transport_ticket: ticket 78.7% of class, 81% of its 169 rows; iot_cleaning: vacuum 72.9% of class, 100% of its 86 rows; qa_stock: stock 71.1% of class, 97% of its 144 rows; general_joke: joke 70.7% of class, 96% of its 68 rows; play_podcasts: podcast 65.1% of class, 99% of its 177 rows; add_to_playlist: playlist 63.6% of class, 81% of its 1336 rows; transport_ticket: train ticket 62.6% of class, 81% of its 134 rows; general_confirm: confirm 60.0% of class, 96% of its 273 rows; get_refund: money 59.6% of class, 73% of its 582 rows; alarm_query: alarms 57.5% of class, 81% of its 104 rows; play_radio: radio 55.7% of class, 89% of its 248 rows; review: feedback 54.9% of class, 100% of its 439 rows; complaint: against 54.6% of class, 98% of its 439 rows; rate_book: 6 54.0% of class, 92% of its 972 rows; news_query: news 53.9% of class, 81% of its 452 rows; transport_taxi: taxi 53.3% of class, 99% of its 74 rows; rate_book: of 6 50.8% of class, 99% of its 846 rows; rate_book: rate 50.5% of class, 85% of its 979 rows; transport_ticket: a train 50.0% of class, 76% of its 115 rows; delivery_period: arrive 48.3% of class, 95% of its 396 rows; search_screening_event: movie 48.3% of class, 85% of its 910 rows; email_query: emails 47.2% of class, 96% of its 279 rows; social_post: tweet 46.9% of class, 98% of its 189 rows; qa_currency: exchange 46.5% of class, 97% of its 136 rows; book_restaurant: book a 46.3% of class, 90% of its 893 rows; transport_ticket: a train ticket 44.3% of class, 81% of its 95 rows; add_to_playlist: to my 41.2% of class, 75% of its 930 rows; iot_cleaning: cleaner 40.7% of class, 98% of its 49 rows; recover_password: my user 40.5% of class, 80% of its 396 rows; check_cancellation_fee: early 40.1% of class, 97% of its 308 rows; complaint: claim 40.0% of class, 100% of its 316 rows; transport_taxi: a taxi 39.4% of class, 100% of its 54 rows; email_sendemail: email to 39.0% of class, 91% of its 216 rows; alarm_set: alarm for 38.5% of class, 84% of its 101 rows

Shortcut models

Predicting the label class on test (5,577 rows). Chance 1.0%, majority class ('play_music') 3.9%; balanced chance 1.0%.

model (logistic regression, trained on the train sample)	test accuracy	balanced accuracy (mean recall)
bag of words, whole state (words and word pairs)	94.1%	90.0%
bag of words, main text only (`message`)	78.7%	66.4%
surface features only (no words: length, punctuation, case, markup, digits, script, state sizes, option count, instruction kind)	21.2%	13.3%
surface features of the main text only	17.5%	10.7%

Strongest single surface features (logistic regression on one feature, balanced accuracy on test):

feature	accuracy	balanced accuracy
state_chars(log)	4.6%	2.3%
other_state_chars(log)	4.5%	2.1%
n_options(log)	4.5%	2.1%
all_lower	4.4%	2.0%
digit_ratio	5.2%	2.0%
ends_.	5.0%	1.9%
ends_other	5.0%	1.8%
count_.	4.9%	1.8%
count_-	4.9%	1.8%
count_,	4.9%	1.8%

Other state fields alone (predicting the label class on test from one field, without the main text):

field	treated as	accuracy	balanced accuracy
service	categorical, 3 values	6.2%	3.1%

No-meaning option picker: a logistic ranker scores each option from its position, length, key type, fixed-option identity and shape (commas, brackets, capitals), never reading the state or the option's words, and picks the top option per row.

Test accuracy 5.5% against uniform chance 4.1% (this includes the fixed options, whose share is a class prior).

2. Duplicates and split separation

Families shared between splits

splits	shared families	examples
train ∩ dev	0
train ∩ calibration	0
train ∩ test	0
dev ∩ calibration	0
dev ∩ test	0
calibration ∩ test	0

Train rows whose main text repeats an earlier row's (normalised): 75 (0.1%); groups: 74; largest group 3.
Train rows identical in the whole prompt (state, options, instructions): 32.
Identical whole prompt, different answer: 0 groups (0 rows).
Identical main text, different label class: 4 groups (8 rows). (Can be legitimate when the rest of the state or the options differ.)
- "play pandora" ×2: play_radio 1, play_music 1
- "contact customer service" ×2: social_post 1, contact_customer_service 1
- "show me movie times" ×2: search_screening_event 1, recommendation_movies 1
- "complaint" ×2: social_post 1, complaint 1

Most repeated main texts in train:

×3: "editing details on standard account" (edit_account 3)
×2: "what do i need to do to switch to the standard account?" (switch_account 2)
×2: "what do i ned to do to change to the business account?" (switch_account 2)
×2: "use pro account" (switch_account 2)
×2: "updating pro accojnt" (edit_account 2)
×2: "updating info on premium account" (edit_account 2)
×2: "updating data on pro account" (edit_account 2)
×2: "update information on platinum account" (edit_account 2)

Main text of held-out rows found verbatim in train (normalised; the leak gate ignores short texts shared by many items):

split	rows	examples
dev	0 (0.0%)
calibration	0 (0.0%)
test	2 (0.0%)	"play spotify"; "what is the weather here"

Near duplicates (MinHash, word 3-gram Jaccard ≥ 0.8 on the main text)

Train: 1,482 near-duplicate pairs; 2,493 rows (4.5%) sit in 1,111 clusters; largest cluster 9; excess rows (cluster size − 1) 1,382 (2.5%).
Clusters with more than one label class: 15 (33 rows).
- ×3: "using freemium accouny" → switch_account 2, delete_account 1
- ×3: "using premium accouny" → switch_account 2, delete_account 1
- ×3: "using standard accouny" → switch_account 2, delete_account 1
- ×2: "ah it's amazing." → general_praise 1, general_affirm 1
- ×2: "play pandora" → play_radio 1, play_music 1
Held-out rows with a near duplicate in train: dev 0 (0.0%), calibration 0 (0.0%), test 7 (0.1%)
- train "modifying standard account" ~ test "mldifying standard account" (J=0.83)
- train "what is the weather here" ~ test "what is the weather here" (J=1.00)
- train "i have a problem with creating a platinum account" ~ test "have a problem with creating a platinum account" (J=0.86)
- train "What movies are playing nearby?" ~ test "what movies are playing nearby" (J=1.00)
- train "play Spotify" ~ test "play spotify" (J=1.00)

Largest train clusters:

×9: "dammit, you gave me not correct information."
×8: "where to see at what time I can contact customer service?"
×7: "can you show me how soon can I expect my delivery?"
×6: "that does not really matter to me."
×6: "see in what situations can I ask to be refunded"

5. Junk

split	empty main text	main text under 10 characters
train	0	255
dev	0	3
calibration	0	7
test	0	29

Very short train examples: "unmute" (audio_volume_up); "swimming" (recommendation_events); "quieter" (audio_volume_down); "hail taxi" (transport_taxi); "remind" (calendar_set); "wine shop" (recommendation_locations); "tv" (general_quirky); "porn" (general_quirky); "reminders" (calendar_set); "investing" (general_quirky); "meetings" (calendar_set); "open fm" (play_radio)

Pattern scan of train main texts (count, then the share of each class's rows):

pattern	rows	by class
placeholder [NAME]-style	0 (0.0%)
lorem ipsum	0 (0.0%)
TODO/TBD/FIXME	6 (0.0%)	add_to_playlist 0.3%, play_music 0.0%
'As an AI' / refusal	0 (0.0%)
chat preamble ('Here is/are...', 'Sure!')	3 (0.0%)	email_addcontact 1.4%, email_sendemail 0.2%, general_quirky 0.1%
meta words (example/variation/message:)	0 (0.0%)
model thinking tags	0 (0.0%)
JSON/code-fence leftovers	0 (0.0%)
encoding garbage (mojibake/replacement char)	0 (0.0%)
HTML tag	0 (0.0%)
HTML entity	0 (0.0%)
base64-like run (40+ chars)	0 (0.0%)
URL	0 (0.0%)

TODO/TBD/FIXME: support-intents:snips:train:PlayMusic-train-1410:intent (play_music): Please play playlist TODO REGGAETON | support-intents:snips:train:AddToPlaylist-train-1280:intent (add_to_playlist): Add artist to TODO EDM | support-intents:snips:train:AddToPlaylist-train-1099:intent (add_to_playlist): Add track by Fleetwood Mac to lakeisha's playlist TODO NOVEDADES
chat preamble ('Here is/are...', 'Sure!'): support-intents:hwu64:none:hwu64-19181:intent (email_sendemail): here is a new email address to be added to the contacts | support-intents:hwu64:none:hwu64-19576:intent (email_addcontact): here's an email address | support-intents:hwu64:none:hwu64-7491:intent (general_quirky): here is something from today

Repeated sentences across rows (≥25 characters, in at least 0.2% of the sample):

none

6. Samples

20 random train rows per kind: support-intents-samples.txt. Reading notes are in the findings above.