Providing a self-maintaining automated chat response generator

• US 10,394,853 B2
• Filed: 08/21/2017
• Issued: 08/27/2019
• Est. Priority Date: 08/21/2017
• Status: Active Grant

• Current Assignee: Qualtrics LLC (SAP SE)
• Original Assignee: Qualtrics LLC (Australia) (SAP SE)
• Inventors: Porter, Michael
• Primary Examiner(s): Riley, Marcus T
• Application Number:
  US15/681,858
• Publication Number:
  US 20190057143A1
• Time in Patent Office: 736 Days
• Field of Search: None
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06N 3/02 Neural networks
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06N 5/046 Forward inferencing; Production systems
• CPC Class Codes: G06Q 10/063114 Status monitoring or status determination for a person or group
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 30/016 After-sales
• CPC Class Codes: H04L 12/1822 Conducting the conference, e.g. admission, detection, selection or grouping of participants, correlating users to one or more conference sessions, prioritising transmission
• CPC Class Codes: H04L 51/04 Real-time or near real-time messaging, e.g. instant messaging [IM]

Abstract

The present disclosure is directed toward systems and methods for providing a self-maintaining automated chat response generator. In particular, the systems and methods described herein analyze a corpus of digital content to identify content topics and generate a language model for categorizing a chat question. Additionally, the systems and methods described herein analyze a chat question to assign the chat question to a content topic based on keywords identified within the chat question. The systems and methods also generate a response to provide to the chat question.

Claims

1. A method comprising:

analyzing, by at least one processor of a digital communication response system, a digital content body comprising a plurality of content portions to identify, from within the plurality of content portions, content terms corresponding to a plurality of content topics;

analyzing, by the at least one processor, a chat history comprising a plurality of chat messages between one or more client devices and the digital communication response system to identify, from within the plurality of chat messages, chat terms corresponding to the plurality of content topics;

determining relational rules between the content terms, between the chat terms, and between the content terms and the chat terms;

generating a language model to map the relational rules for generating responses to chat questions;

monitoring the digital content body and the chat history to detect a change within at least one of the plurality of content portions of the digital content body or a change within the chat history; and

in response to detecting the change within the at least one of the plurality of content portions or the change within the chat history, generating an updated language model that comprises one or more updated relational rules between the content terms, between the chat terms, and between the content terms and the chat terms.

2. The method of claim 1, further comprising:

receiving a chat question originating from a client device;

determining a response to the chat question by;

analyzing the received chat question to identify keywords within the chat question;

applying the updated language model to the keywords to determine, in accordance with the relational rules, one or more relationships between the keywords, the content terms, and the chat terms; and

categorizing the chat question to at least one content topic from among the plurality of content topics in accordance with the one or more relationships; and

providing, for presentation on the client device, the response to the chat question.

3. The method of claim 2, further comprising:

determining, for each of the plurality of content topics, a relevance to the chat question; and

identifying the at least one content topic from among the plurality of content topics as relevant to the chat question by comparing the relevance of each of the plurality of content topics.

4. The method of claim 3, wherein determining the relevance of each of the plurality of content topics to the chat question comprises analyzing the one or more relationships between the keywords and the content terms corresponding to each of the plurality of content topics.

5. The method of claim 3, wherein identifying the at least one content topic as relevant to the chat question comprises determining that the at least one content topic has a relevance above a relevance threshold.

6. The method of claim 3, wherein providing the response to the chat question comprises sending, to the client device, a content portion from among the plurality of content portions that corresponds to the identified at least one content topic.

7. The method of claim 1, wherein generating the updated language model comprises adding a new content term corresponding to the plurality of content topics in response to detecting the new content term within the digital content body based on monitoring the digital content body.

8. The method of claim 1, wherein generating the updated language model comprises modifying one or more content terms corresponding to the plurality of content topics in response to detecting a change to the one or more content terms within the digital content body.

9. The method of claim 1, wherein generating the updated language model comprises modifying one or more of the plurality of content topics in response to detecting the change within the at least one of the plurality of content portions.

10. The method of claim 1, wherein generating the updated language model is based on detecting a new chat term within the chat history.

11. The method of claim 10, further comprising updating the relational rules based on the new chat term.

12. The method of claim 1, wherein detecting the change within the at least one of the plurality of content portions comprises detecting, from a content provider, one or more of:

an addition of content, a deletion of content, or an alteration of content.

13. A non-transitory computer readable medium comprising instructions that, when executed by at least one processor, cause a computer device to:

analyze a digital content body comprising a plurality of content portions to identify, from within the plurality of content portions, content terms corresponding to a plurality of content topics;

analyze a chat history comprising a plurality of chat messages from one or more client devices to identify, from within the plurality of chat messages, chat terms corresponding to the plurality of content topics;

determine relational rules between the content terms, between the chat terms, and between the content terms and the chat terms;

generate a language model to map the relational rules for generating responses to chat questions;

monitor the digital content body and the chat history to detect a change within at least one of the plurality of content portions of the digital content body or a change within the chat history; and

in response to detecting the change within the at least one of the plurality of content portions or the change within the chat history, generate an updated language model that comprises one or more updated relational rules between the content terms, between the chat terms, and between the content terms and the chat terms.

14. The non-transitory computer readable medium of claim 13, wherein the instructions cause the computer device to detect the change within the chat history by detecting a new entry to the chat history.

15. The non-transitory computer readable medium of claim 13, further comprising instructions that, when executed by the at least one processor, cause the computer device to:

identify one or more keywords within a chat question received from the client device; and

determine one or more relationships between the keywords, the content terms, and the chat terms.

16. The non-transitory computer readable medium of claim 13, wherein the instructions cause the computer device to determine the relational rules by comparing vector representations of the content terms and the chat terms in a vector space.

17. The non-transitory computer readable medium of claim 16, further comprising instructions that, when executed by the at least one processor, cause the computer device to:

determine, in response to receiving a chat question from a client device, a most relevant content topic for the chat question based on applying the updated language model to the chat question; and

provide, to the client device, a content portion of the digital content body that corresponds to the most relevant content topic.

18. The non-transitory computer readable medium of claim 17, wherein the instructions cause the computer device to apply the updated language model to the chat question by implementing a word mover'"'"'s distance algorithm to determine in vector space a distance between a vector representation of the chat question and vector representations of each of the plurality of content topics.

19. The non-transitory computer readable medium of claim 18, further comprising instructions that, when executed by the at least one processor, cause the computer device to:

generate, based on the content portion that corresponds to the most relevant content topic, a natural language response to the chat question; and

provide, to the client device, the natural language response.

20. A system comprising:

at least one processor; and

a non-transitory computer readable medium comprising instructions thereon that, when executed by the at least one processor, cause the system to;

analyze a digital content body comprising a plurality of content portions to identify, from within the plurality of content portions, content terms corresponding to a plurality of content topics;

analyze a chat history comprising a plurality of chat messages from one or more client devices to identify, from within the plurality of chat messages, chat terms corresponding to the plurality of content topics;

determine relational rules between the content terms, between the chat terms, and between the content terms and the chat terms;

generate a language model to map the relational rules for generating responses to chat questions;

monitor the digital content body and the chat history to detect a change within at least one of the plurality of content portions of the digital content body or a change within the chat history; and

in response to detecting the change within the at least one of the plurality of content portions or the change within the chat history, generate an updated language model that comprises one or more updated relational rules between the content terms, between the chat terms, and between the content terms and the chat terms;

receive a chat question originating from a client device; and

determine a response to the chat question based on the updated language model.