Context-sensitive generation of conversational responses

Patent metadata

Field	Value
Publication number	US-10536402-B2
Application number	US-201816112611-A
Country	US
Kind code	B2
Filing date	Aug 24, 2018
Priority date	May 31, 2015
Publication date	Jan 14, 2020
Grant date	Jan 14, 2020

Abstract

Examples are generally directed towards context-sensitive generation of conversational responses. Context-message-response n-tuples are extracted from at least one source of conversational data to generate a set of training context-message-response n-tuples. A response generation engine is trained on the set of training context-message-response n-tuples. The trained response generation engine automatically generates a context-sensitive response based on a user generated input message and conversational context data. A digital assistant utilizes the trained response generation engine to generate context-sensitive, natural language responses that are pertinent to user queries.

First claim

What is claimed is:

1. A system for generating context-sensitive conversational responses, said system comprising:
at least one processor;
a memory storage device, the memory storage device comprising a memory area storing a response generation engine, wherein the at least one processor executes the response generation engine to:
receive a user generated message associated with a current conversation;
derive conversational context data associated with the user generated message, wherein the derived conversational context data is based at least on a set of context data sources associated with a user, the set of context data sources comprising:
linguistic context data captured as message-response pairs of a conversation of the user, and
non-linguistic context data associated with a physical or virtual environment of the user; and

output a generated context-sensitive response to the user generated message based at least on the conversational context data and the user generated message.

2. The system of claim 1 , further comprising:
an extraction component, wherein the at least one processor further executes the extraction component to:
select context-message-response n-tuples corresponding to a selected context-message data pair, wherein the context-message-response n-tuples are extracted from at least one source of conversational data, wherein a context-message-response n-tuple comprises a human-generated message and a reference response corresponding to the human-generated message; and
train the response generation engine in a context-sensitive manner using the selected context-message-response n-tuples to form a trained response generation engine.

3. The system of claim 2 , wherein the at least one processor further executes the extraction component to:
extract the context-message-response n-tuples from at least one social media source, wherein the at least one social media source provides the conversational context data in at least one format, wherein a format of the conversational data comprises at least one of a text format, an audio format, or a visual format.

4. The system of claim 1 , further comprising:
a context data derivation component, wherein the at least one processor further executes the context data derivation component to:
derive the linguistic context data from conversation history data, wherein the conversation history data comprises prior message and response pairs occurring within the current conversation.

5. The system of claim 1 , further comprising:
a context data derivation component, wherein the at least one processor further executes the context data derivation component to:
derive non-linguistic context data from sensor data received in real-time from a set of sensors, wherein the set of sensors comprises at least one of a camera, an audio sensor, a global positioning system (GPS) sensor, an infrared sensor, a pressure sensor, a motion sensor, an orientation sensor, temperature sensor, or speed sensor, wherein the context data derivation component provides the conversational context data to the response generation engine.

6. The system of claim 1 further comprising:
a context data derivation component, wherein the at least one processor further executes the context data derivation component to:
access the non-linguistic context data from a set of user data, the non-linguistic context data comprising at least one of preferences data, contacts data, health data, interest data, user activity data, calendar data, activities data, work data, hobby data, or daily routine data.

7. The system of claim 1 , wherein the response generation engine comprises a machine learning model.

8. One or more computer storage media having computer-executable components, said components comprising:
an extraction component that is executed to cause at least one processor to extract context-message-response n-tuples from at least one source of conversational data based on a context-message pair to form a training set of context-message-response n-tuples;
a response generation training component that is executed to cause at least one processor to train a response generation engine using the training set of context-message-response n-tuples to form a trained response generation engine; and
the trained response generation engine that is executed to cause at least one processor to generate a context-sensitive conversational response to a user generated message based on conversational context data associated with a user, the conversational context data being gathered during a current conversation, wherein the conversational context data is based at least on a set of context data sources associated with a user, the set of context data sources comprising:
linguistic context data captured as context-message-response n-tuples of a conversation of the user, and
non-linguistic context data associated with a physical or virtual environment of the user.

9. The computer storage media of claim 8 , wherein the at least one source of conversational data comprises a social media source, wherein the social media source provides conversational data in at least one format, wherein a format of the conversational data comprises at least one of a text format, an audio format, or a visual format.

10. The computer storage media of claim 8 , wherein the response generation engine comprises a machine learning model.

11. A computer-implemented method for state-free generation of context-sensitive conversational responses, said method comprising:
receiving, by a response generation engine implemented on a processor, a user generated message associated with a current conversation;
deriving, by a context data derivation component implemented on the processor, conversational context data associated with the user generated message, wherein the derived conversational context data is based at least on a set of context data sources associated with a user, the set of context data sources comprising:
linguistic context data captured as message-response pairs of a conversation of the user, and
non-linguistic context data associated with a physical or virtual environment of-the user; and
outputting a generated context-sensitive response to the user generated message based at least on the conversational context data and the user generated message.

12. The computer-implemented method of claim 11 , further comprising:
selecting, by an extraction component, context-message-response n-tuples corresponding to a selected context-message data pair, wherein the context-message-response n-tuples are extracted from at least one source of conversational data, wherein a context-message-response n-tuple comprises a human-generated message and a reference response corresponding to the human-generated message; and
training the response generation engine in a context-sensitive manner using the selected context-message-response n-tuples to form a trained response generation engine.

13. The computer-implemented method of claim 12 , wherein the at least one source of conversational data comprises a social media source, wherein the social media source provides conversational data in at least one format, wherein a format of the conversational data comprises at least one of a text format, an audio format, or a visual format.

14. The computer-implemented method of claim 12 , further comprising:
extracting, by the extraction component, the context-message-response n-tuples from at least one social media source, wherein the at least one social media source provides the conversational context data in at least one format, wherein a format of the conversational

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Galley Michel
• Sordoni Alessandro
• Brockett Christopher John
• Gao Jianfeng
• Dolan William Brennan
• Ji Yangfeng
• Auli Michael
• Mitchell Margaret Ann
• Nie Jian-Yun

Classifications

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06F40/56 Primary

  Natural language generation · CPC title

• H04L51/02 Primary

  using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages · CPC title

• G06F17/2881

  Physics · mapped topic