Systems and methods for a multiple topic chat bot

Patent metadata

Field	Value
Publication number	US-10366168-B2
Application number	US-201715404932-A
Country	US
Kind code	B2
Filing date	Jan 12, 2017
Priority date	Jan 12, 2017
Publication date	Jul 30, 2019
Grant date	Jul 30, 2019

Abstract

Systems and methods for multiple topic automated chatting are provided. The systems and method provide multiple topic automated (or artificial intelligence) chatting by analyzing user inputs in a conversation to determine a plurality topics, to determine and score features related to the determined topics and different users, and to create a knowledge graph of the determined topics. Based on these determinations, the systems and methods may determine if a reply should be provided and then predict a reply.

First claim

The invention claimed is:

1. A system for a multiple topic chat bot, the system comprising:
at least one processor; and
a memory for storing and encoding computer executable instructions that, when executed by the at least one processor is operative to:
collect user inputs in a conversation to form a collection;
analyze the collection to determine topics in the conversation;
assign an emotion label to each topic;
identify a relationship between different users;
score a closeness of the relationship based on social connection, agreement, and sentiment analysis to form a scored first feature;
score each user's interest in each topic based on user sentiment toward each topic and engagement frequency in each topic to form a scored second feature;
score an engagement rate for each topic of the topics based on a number of users engaged in a topic, frequency of the topic in the conversation, timing of the topic, and the user sentiment toward the topic to form a scored third feature;
create a knowledge graph of the topics that graphs relationships between the topics utilizing topic keywords based on the collection and world knowledge;
determine that a first topic meets a relevancy threshold based on scored features for the first topic, wherein the scored features include the scored first feature, the scored second feature, and the scored third feature for the first topic;
predict, utilizing a trained model, one or more first responses based on the knowledge graph and the user inputs associated with the first topic;
provide the one or more first responses to the conversation;
predict one or more second responses utilizing the knowledge graph and the user inputs associated with a second topic; and
provide the one or more second responses to the conversation.

2. The system of claim 1 , wherein the user inputs do not include a query.

3. The system of claim 1 , wherein the at least one processor is operative to:
identify that a query included in the user inputs is associated with the first topic, wherein the first topic meets the relevancy threshold because the first topic includes the query, and
wherein the one or more first responses is a reply to the query.

4. The system of claim 1 , wherein the scored first feature, the scored second feature, and the scored third feature each utilize a same scoring scale.

5. The system of claim 1 , wherein the at least one processor is operative to: score the closeness of the relationship between different users utilizing a first learning algorithm;
score each user's interest in each topic utilizing a second learning algorithm; and
score the engagement rate for each topic utilizing a third learning algorithm.

6. The system of claim 5 , wherein the at least one processor is operative to:
collect at least one of user feedback and world feedback; and
train at least one of the first learning algorithm, the second learning algorithm, and the third learning algorithm based on the at least one of the user feedback and the world feedback.

7. The system of claim 1 , wherein the user inputs are from at least a first user and second user.

8. The system of claim 1 , wherein the at least one processor is operative to:
compare each topic of the topics to a timing threshold, wherein the topics include the first topic, a second topic, and a third topic;
determine that the third topic breaches the timing threshold; and
delete the third topic in response to determining that the third topic breaches the timing threshold.

9. The system of claim 1 , wherein utilizing the trained model comprises:
utilizing a learning-to-rank architecture of pairwise learning for constructing a relevance-based response ranking model.

10. The system of claim 9 , wherein the at least one processor is operative to utilize a gradient boosting decision tree for the relevance-based response ranking model to predict the one or more first responses utilizing the knowledge graph for the first topic.

11. The system of claim 1 , wherein the one or more first responses is a summary of opinions for the first topic.

12. The system of claim 1 , wherein the one or more first responses is a new topic.

13. The system of claim 1 , wherein the at least one processor is operative to
rank each topic in the topics based on the scored first feature, the scored second feature, and the scored third feature for each topic.

14. The system of claim 1 , wherein the at least one processor is operative to
determine that a second topic meets the relevancy threshold.

15. A method for emotionally intelligent automated chatting, the method comprising:
collecting inputs in a conversation to form a collection;
analyzing the collection to determine topics in the conversation;
assign a sentiment to each topic;
scoring an engagement rate for each topic to form an engagement score for each topic;
scoring a user interest in each topic to form an interest score for each topic;
creating a knowledge graph between the topics that graphs relationships between the topics;
determining a relationship between each set of users in the conversation;
scoring a closeness of the relationship to form a closeness score for each relationship;
determining that a first topic of the topics meets a relevancy threshold based on the engagement score, the interest score of the first topic, and the closeness score between each set of users engaged in the first topic;
predicting, utilizing a trained model, a first response based on the knowledge graph and inputs associated with the first topic;
providing the first response to the conversation;
predicting one or more second responses utilizing the knowledge graph and inputs associated with a second topic; and
providing the one or more second responses to the conversation.

16. The method of claim 15 , further comprising
identifying that a query included in the inputs is associated with the first topic,
wherein the first topic meets the relevancy threshold because the first topic includes the query, and
wherein the first response is a reply to the query.

17. The method of claim 15 , wherein the inputs do not include a query, and
wherein the engagement score and the interest score are averaged before comparison to the relevancy threshold.

18. The method of claim 15 , further comprising:
comparing each topic of the topics to a timing threshold, wherein the topics includes the first topic, a second topic, and a third topic;
determining that that the third topic breaches the timing threshold; and
deleting the third topic in response to determining that the third topic breaches the timing threshold.

19. The method of claim 15 , wherein the conversation includes at least two users.

20. A system for a multiple topic chat bot, the system comprising:
at least one processor; and
a memory for storing and encoding computer executable instructions that, when executed by the at least one processor is operative to:
collect user inputs from a group chat of a first user and a second user to form a collection;
analyze the collection to determine a first topic and a second topic;
assign a sentiment to each of the first topic and the second topic;
create a knowledge graph of the first topic and the second topic;
identify a first relationship between the first user and the second user;
score a closeness of the first relationship to form a scored first relationship;
score an interest of each of the first user and the second user in the first topic to form a scored first user-first

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Wu Xianchao

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F40/56

  Natural language generation · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F16/9024

  Graphs; Linked lists (G06F16/9027 takes precedence) · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title