IP Library › Granted Patent US 12,592,234
Granted Patent B1
US 12,592,234 · App. 18/229,837 · Granted Mar 31, 2026

Natural language processing policies

Inventors: Arunachalam Sundararaman (Redmond, WA); Balasubramaniam K Thandu (Sammamish, WA); Rajesh Ravindran Nandyaleth (Bellevue, WA); Hermilo H Cardenas Juarez (Seattle, WA); Ananth Duggirala (Redmond, WA)
Assignee: Amazon Technologies, Inc.
G10L15/22G10L15/1815G10L2015/223
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,592,234
App. No.
18/229,837
Filed
Aug 3, 2023
Granted
Mar 31, 2026
Kind
B1
Art Unit
2654
USPC
704/257
Abstract

Systems and methods for natural language processing policies are disclosed. For example, content such as a response to a user command and/or user input such as a customer review, may be analyzed to determine if the content deviates from one or more content policies and to determine a severity of the content-policy deviation(s). A system may determine a number of content-policy deviations associated with the content and whether the content includes an indicator of a person, such as a pronoun and/or a proper noun. Content that includes multiple content-policy deviations and/or that includes an indicator of a person may be determined to be more severe than other deviated content. Applications associated with deviated content may be ranked and corrective action may be taken based on the ranking.

Claims (62)

1 . A system comprising:

one or more processors; and

non-transitory computer-readable media storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:

selecting a content policy, from among multiple content policies, to apply to content received in association with account data;

identifying a condition associated with the content policy by determining the content is associated with a predefined maturity rating, wherein the condition is associated with the predefined maturity rating;

receiving first data representing the content, the first data received based at least in part on request data sent from a device associated with the account data;

determining, based on the first data, that a phrase of the content violates the content policy;

detecting the condition from the first data; and

causing an action to be performed in association with the content based at least in part on the phrase violating the content policy and detecting the condition.

2 . The system of claim 1 , wherein:

the condition comprises the content having a predefined data type; and

detecting the condition comprises detecting that the content is of the predefined data type.

3 . The system of claim 1 , wherein:

the request data comprises audio data representing a voice command; and

causing the action to be performed comprises causing the action to be performed based at least in part on the request data comprising the audio data.

4 . The system of claim 1 , wherein:

the condition comprises receiving the first data from a predefined application; and

detecting the condition comprises detecting that the first data was received from the predefined application.

5 . The system of claim 1 , wherein:

the condition comprises the device having a predefined device type identified by the content policy; and

detecting the condition comprises determining that the device is associated with the predefined device type.

6 . The system of claim 1 , wherein:

the condition comprises receiving the request data from one or more devices associated with a predefined user profile of the account data; and

detecting the condition comprises determining that the request data was received from the device associated with the predefined user profile.

7 . The system of claim 1 , the operations further comprising determining the action to be performed based at least in part on the request data, and wherein:

the condition comprises the action being of a predefined action type; and

detecting the condition comprises determining that the action is of the predefined action type.

8 . The system of claim 1 , wherein:

the content policy comprises a first content policy of the multiple content policies;

the first content policy indicates that the action is to filter the content; and

a second content policy of the multiple content policies indicates that the action is to provide the content unfiltered.

9 . The system of claim 1 , the operations further comprising determining, based at least in part on the content policy, a content policy violation severity value associated with the content.

10 . A method comprising:

selecting a content policy of multiple content policies to apply to content received in association with account data;

identifying a condition associated with the content policy by determining the content is associated with a predefined maturity rating, wherein the condition is associated with the predefined maturity rating;

receiving first data representing the content, the first data received based at least in part on request data sent from a device associated with the account data;

determining that a phrase of the content violates the content policy;

detecting the condition from the first data; and

causing an action to be performed in association with the content based at least in part on the phrase violating the content policy and detecting the condition.

11 . The method of claim 10 , wherein:

the condition comprises the content having a predefined data type; and

detecting the condition comprises detecting that the content is of the predefined data type.

12 . The method of claim 10 , wherein:

the request data comprises audio data representing a voice command; and

causing the action to be performed comprises causing the action to be performed based at least in part on the request data comprising the audio data.

13 . The method of claim 10 , wherein:

the condition comprises receiving the first data from a predefined application; and

detecting the condition comprises detecting that the first data was received from the predefined application.

14 . The method of claim 10 , wherein:

the condition comprises the device having a predefined device type associated with the content policy; and

detecting the condition comprises determining that the device is associated with the predefined device type.

15 . The method of claim 10 , wherein:

the condition comprises receiving the request data from one or more devices associated with a predefined user profile of the account data; and

detecting the condition comprises determining that the request data was received from the device associated with the predefined user profile.

16 . The method of claim 10 , further comprising determining the action to be performed based at least in part on the request data, and wherein:

the condition comprises the action being of a predefined action type; and

detecting the condition comprises determining that the action is of the predefined action type.

17 . The method of claim 10 , wherein:

the content policy comprises a first content policy of the multiple content policies;

the first content policy indicates that the action is to filter the content; and

a second content policy of the multiple content policies indicates that the action is to provide the content unfiltered.

18 . The method of claim 10 , further comprising determining, based at least in part on the content policy, a content policy violation severity value associated with the content.

Continuity (2)
Continuation 17470241 · Sep 9, 2021
Continuation 16574927 · Sep 18, 2019
References Cited (43)
US 6519571B1 · Guheen · 2003 [cited by examiner]
US 8032409B1 · Mikurak · 2011 [cited by examiner]
US 9413891B2 · Dwyer · 2016 [cited by examiner]
US 9992642B1 · Rapp · 2018 [cited by examiner]
US 10044570B2 · Hinrichs · 2018 [cited by examiner]
US 10237137B2 · Helvey · 2019 [cited by examiner]
US 10645224B2 · Dwyer · 2020 [cited by examiner]
US 10706848B1 · Greene · 2020 [cited by examiner]
US 10833947B2 · Helvey · 2020 [cited by examiner]
US 10978056B1 · Challa · 2021 [cited by examiner]
US 11089081B1 · Karppanen · 2021 [cited by examiner]
US 11102624B2 · Rapp · 2021 [cited by examiner]
US 11120799B1 · Sundararaman · 2021 [cited by examiner]
US 11188376B1 · Alexander · 2021 [cited by examiner]
US 11289075B1 · Pandey · 2022 [cited by examiner]
US 11423451B1 · Chaudhari · 2022 [cited by examiner]
US 11544152B2 · Savir · 2023 [cited by examiner]
US 11551681B1 · Pandey · 2023 [cited by examiner]
US 11663182B2 · Emma · 2023 [cited by examiner]
US 11763816B1 · Sundararaman · 2023 [cited by examiner]
US 12327562B1 · Bodigutla · 2025 [cited by examiner]
US 20040064351A1 · Mikurak · 2004 [cited by examiner]
US 20070156269A1 · Suryanaraya · 2007 [cited by examiner]
US 20090070103A1 · Beggelman · 2009 [cited by examiner]
US 20100010968A1 · Redlich · 2010 [cited by examiner]
US 20100145700A1 · Kennewick · 2010 [cited by examiner]
US 20100211431A1 · Lutnick · 2010 [cited by examiner]
US 20130253910A1 · Turner · 2013 [cited by examiner]
US 20150309983A1 · Hoover · 2015 [cited by examiner]
US 20160057026A1 · Hinrichs · 2016 [cited by examiner]
US 20160147731A1 · Parikh · 2016 [cited by examiner]
US 20160292445A1 · Lindemann · 2016 [cited by examiner]
US 20160359684A1 · Rizqi · 2016 [cited by examiner]
US 20170154188A1 · Meier · 2017 [cited by examiner]
US 20170206064A1 · Breazeal · 2017 [cited by examiner]
US 20170243028A1 · LaFever · 2017 [cited by examiner]
US 20170300472A1 · Parikh · 2017 [cited by examiner]
US 20180077025A1 · Helvey · 2018 [cited by examiner]
US 20180165554A1 · Zhang · 2018 [cited by examiner]
US 20180310144A1 · Rapp · 2018 [cited by examiner]
US 20200076696A1 · Helvey · 2020 [cited by examiner]
US 20220399023A1 · Gyanchandani · 2022 [cited by examiner]
Office Action for U.S. Appl. No. 17/470,241, mailed on Jan. 30, 2023, Arunachalam Sundararaman, Natural Language Processing Policies , 12 pages. [cited by applicant]