Implicit dialog approach operating a conversational access interface to web content

Patent metadata

Field	Value
Publication number	US-10915588-B2
Application number	US-201816053726-A
Country	US
Kind code	B2
Filing date	Aug 2, 2018
Priority date	Aug 2, 2018
Publication date	Feb 9, 2021
Grant date	Feb 9, 2021

Abstract

A method, apparatus and computer program product for presenting a user interface for a conversational system is described. A user input is received in a dialog between a user and the conversational system, the user input in a natural language. A domain trained semantic matcher is used to determine a set of entities and a user intent from the user input. One or more queries is generated to selected ones of a plurality of knowledge sources, the knowledge sources created from domain specific knowledge. The results from the one or more queries are ranked based on domain specific knowledge. A system response is presented in the dialog based on at least a highest ranked result from the plurality of knowledge sources.

First claim

What is claimed:

1. A method for presenting a user interface for a conversational system comprising:
receiving a user input produced in a dialog between a user and the conversational system, the user input in a natural language;
using a domain trained semantic matcher to determine a set of entities and a user intent from the user input, the domain trained semantic matcher trained according to tasks of a target web site;
generating a plurality of queries from the set of entities and the user intent from the user input, each of the queries to selected ones of a plurality of knowledge sources, the knowledge sources created from domain specific knowledge, the domain specific knowledge related to tasks of the target web site;
ranking results from the plurality of queries based on the domain specific knowledge; and
presenting a system response in the dialog based on at least a highest ranked result from the plurality of knowledge sources.

2. The method as recited in claim 1 , wherein the plurality of queries are generated including a first query to a first knowledge source and a second query to a second knowledge source in the plurality of knowledge sources and a result from the first knowledge source to the first query has a higher confidence score than a result from the second knowledge source to the second query.

3. The method as recited in claim 2 , wherein the system response is generated based on the result from the first knowledge source.

4. The method as recited in claim 1 , wherein the plurality of queries are generated including a first query to a first knowledge source and a second query to a second knowledge source in the plurality of knowledge sources and a result from the first knowledge source to the first query is used to generate a first part of the system response and a result from the second knowledge source to the second query is used to generate a second part of the system response.

5. The method as recited in claim 1 , further comprising generating the system response with a natural language process from the highest ranked result from the plurality of knowledge sources.

6. The method as recited in claim 5 , further comprising:
generating one or more queries to an external knowledge source in addition to the plurality of knowledge sources; and
using a result from the one or more queries to the external knowledge source to generate the system response in the dialog.

7. The method as recited in claim 1 , further comprising updating selected ones of the plurality of knowledge sources based on conversational history in the dialog.

8. The method as recited in claim 1 , wherein the user intent is determined based on conversational history between the user and the conversational system.

9. Apparatus, comprising:
a processor;
computer memory holding computer program instructions executed by the processor for presenting a user interface for a conversational system, the computer program instructions comprising:
program code, operative to receive a user input produced in a dialog between a user and the conversational system, the user input in a natural language;
program code, operative to use a domain trained semantic matcher to determine a set of entities and a user intent from the user input, the domain trained semantic matcher trained according to tasks of a target web site;
program code, operative to generate a plurality of queries from the set of entities and the user intent from the user input, each of the queries to selected ones of a plurality of knowledge sources, the knowledge sources created from domain specific knowledge, the domain specific knowledge related to tasks of the target web site;
program code, operative to rank results from the plurality of queries based on the domain specific knowledge; and
program code, operative to present a system response in the dialog based on at least a highest ranked result from the plurality of knowledge sources.

10. The apparatus as recited in claim 9 , wherein the plurality of queries are generated including a first query to a first knowledge source and a second query to a second knowledge source in the plurality of knowledge sources and a result from the first knowledge source to the first query has a higher confidence score than a result from the second knowledge source to the second query.

11. The apparatus as recited in claim 10 , further comprising program code, operative to generate the system response based on the result from the first knowledge source.

12. The apparatus as recited in claim 9 , wherein the plurality of queries are generated including a first query to a first knowledge source and a second query to a second knowledge source in the plurality of knowledge sources and a result from the first knowledge source to the first query is used to generate a first part of the system response and a result from the second knowledge source to the second query is used to generate a second part of the system response.

13. The apparatus as recited in claim 9 , further comprising program code, operative to generate the system response with a natural language process from the highest ranked result from the plurality of knowledge sources.

14. The apparatus as recited in claim 13 , further comprising:
program code, operative to generate one or more queries to an external knowledge source in addition to the plurality of knowledge sources; and
program code, operative to use a result from the one or more queries to the external knowledge source to generate the system response in the dialog.

15. A computer program product in a non-transitory computer readable medium for use in a data processing system, the computer program product holding computer program instructions executed by the data processing system for presenting a user interface for a conversational system, the computer program instructions comprising:
program code, operative to receive a user input produced in a dialog between a user and the conversational system, the user input in a natural language;
program code, operative to use a domain trained semantic matcher to determine a set of entities and a user intent from the user input, the domain trained semantic matcher trained according to tasks of a target web site;
program code, operative to generate one or more a plurality of queries from the set of entities and the user intent from the user input, each of the queries to selected ones of a plurality of knowledge sources, the knowledge sources created from domain specific knowledge, the domain specific knowledge related to tasks of the target web site;
program code, operative to rank results from the plurality of queries based on the domain specific knowledge; and
program code, operative to present a system response in the dialog based on at least a highest ranked result from the plurality of knowledge sources.

16. The computer program product as recited in claim 15 , wherein the plurality of queries are generated including a first query to a first knowledge source and a second query to a second knowledge source in the plurality of knowledge sources and a result from the first knowledge source to the first query has a higher confidence score than a result from the second knowledge source to the second query.

17. The computer program product as recited in claim 16 , further comprising program code, operative to generate the system response based on the result from the first knowledge source.

18. The computer program product as recited in claim 15 , wherein the plurality of queries are generated including a first query to a first

Assignees

• IBM

Inventors

• Bakis Raimo
• Feng Song
• Ganhotra Jatin
• Gunasekara Chulaka
• Nahamoo David
• Polymenakos Lazaros
• Shashidhara Sunil D
• Wu Cheng
• Zhu Li

Classifications

• G06N3/044

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

• G06N20/00

  Machine learning · CPC title

• G06F40/205

  Parsing · CPC title

• G06F16/9535 Primary

  Search customisation based on user profiles and personalisation · CPC title

• G06F40/14

  Tree-structured documents (parsing G06F40/205; validation G06F40/226) · CPC title