IP Library Granted Patent US 10,380,149
Granted Patent B2
US 10,380,149 · App. 15/503,553 · Granted Aug 13, 2019

Question sentence generating device and computer program

Inventors: Kentaro Torisawa (Tokyo, JP); Jun Goto (Tokyo, JP); Julien Kloetzer (Tokyo, JP); Takuya Kawada (Tokyo, JP)
Assignee: National Institute of Information and Communications Technology
G06F16/3329G06F16/00G06F16/338G06F16/3344
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 10,380,149
App. No.
15/503,553
Granted
Aug 13, 2019
Kind
B2
Abstract

[Object] To provide a device assisting a user to easily generate, in relation to an issue of interest to the user, a question sentence guaranteed to have an answer of a certain accuracy or higher in a question-answering system. [Solution] A question sentence generating device is used with a question-answering system, and it includes: word receiving means for receiving a word 480 as a source for generating a question sentence; and question sentence generating database 502 comprised of a plurality of entries for generating a question sentence. Each of the plurality of entries has a word as a key and includes an answer sentence pattern co-occurring with the word, used in the question-answering system. The question sentence generating device further includes a question sentence generating unit 506 , searching the question sentence generating database 502 for an answer sentence pattern using the word 480 received by the word receiving means as a key, for generating a question sentence from a retrieved answer sentence pattern and the received word 480.

Claims (67)

1. A question sentence generating device for use with a question-answering system, the question sentence generating device comprising:

an interface configured to receive a word or a word sequence as a source for generating a question sentence; and

a memory including a database comprised of a plurality of records for generating question sentences, wherein:

each of said plurality of records has a key word and is generated from an answer having an answer sentence pattern co-occurring with the key word;

a processor configured to:

using the received word as a key, search and retrieve from said question sentence generating database an answer sentence pattern;

generate a question sentence from the retrieved answer sentence pattern and said received word, comprising:

generating question candidates by (1) transforming the retrieved answer sentence pattern into a plurality of question patterns and (2) combining each of the plurality of question patterns with the received word;

scoring each of the question candidates, wherein the score indicates a fit of each question sentence candidate as being a question having as an answer the answer from which the answer sentence pattern is generated; and

selecting as the question sentence the question sentence candidate having a best score; and

output the generated question sentence.

2. The question sentence generating device according to claim 1 , wherein:

the memory additional includes a synonym dictionary configured to store synonymous relation or entailment relation of a word or an answer sentence pattern or both; and

the processor is further configured to:

generate a plurality of question sentences;

select distinct ones of said generated plurality of question sentences to generate a smaller number of question sentences by referring to said synonym dictionary; and

output the generated smaller number of question sentence.

3. The question sentence generating device according to claim 1 , wherein:

the memory additional includes a thesaurus configured to store classification information for classifying words or answer sentence patterns or both systematically in accordance with their meanings; and

the processor is further configured to:

generate a plurality of question sentences;

select distinct ones of said generated plurality of question sentences to a smaller number of question sentences by referring to said thesaurus; and

output the generated smaller number of question sentences.

4. The question sentence generating device according to claim 1 , wherein

said processor is configured to:

receive a plurality of words; and

search said question sentence generating database, using the received plurality of words, for an answer sentence pattern co-occurring with the received plurality of words.

5. The question sentence generating device according to claim 1 , wherein

said processor is further configured to:

search and retrieve said question sentence generating database for an answer sentence pattern using the received word as a key;

generate a question sentence and an answer sentence to the question sentence from the retrieved answer sentence pattern and said received word; and

output the generated question sentence and the generated answer sentence to the question sentence.

6. The question sentence generating device according to claim 1 , wherein

said question-answering system includes a factoid type question-answering sub-system responsive to a factoid type question sentence;

said factoid type question-answering sub-system is configured to generate an answer sentence candidate by searching an answer sentence generating database using a key consisting of a pattern of the question sentence and a word or a word sequence included in the question sentence; and

said answer sentence generating database and said question sentence generating database include common records generated from the same corpus.

7. The question sentence generating device according to claim 1 , wherein:

said question-answering system further includes a non-factoid type question-answering sub-system responsive to a non-factoid type question;

said non-factoid type question-answering sub-system is configured:

to extract a plurality of passages as answer sentence candidates from a corpus,

to calculate a score indicating fitness as an answer to said given question by using a pre-learned scoring means, and

to output an answer sentence candidate having the highest score as an answer to said given question;

the memory further includes a question sentence database configured to store a plurality of question sentences each generated from any of a set of passages prepared in advance and not explicitly distinguished as a question or an answer and having as an answer a passage as a source of said question sentence;

the processor is further configured to:

receive a second word or a second word sequence as a source of generating a question sentence;

generate and output a new question sentence from the received second word or second word sequence as the source of generating the question sentence or synonyms or entailments of these and from a question sentence stored in said question sentence database by referring to said question sentence database; and

collectively output the generated question sentence and the generated new question sentence.

8. The question sentence generating device according to claim 1 , wherein the scoring of each of the question candidates is performed using a machine learning algorithm.

9. A question sentence generating device for use with a question-answering system, wherein:

said question-answering system includes:

a non-factoid type question-answering sub-system responsive to a non-factoid type question, wherein the non-factoid type question-answering sub-system is configured to:

extract a plurality of passages as answer sentence candidates from a corpus,

calculate a score for each passage, said score indicating fitness as an answer to said given question; and

output an answer sentence candidate having the highest score as an answer to said given question;

said question sentence generating device comprising:

a memory including a question sentence database configured to store a plurality of question sentences each generated from any passage in said corpus and having as an answer a passage as a source of generating the question sentence;

an interface configured to receive a word or a word sequence; and

a processor configured to generate and output a new question sentence from the received word or word sequence as the source of generating the question sentence or synonyms or entailments of these and from a question sentence stored in said question sentence database by referring to said question sentence database.

10. A computer program stored on a non-transitory computer readable medium causing a computer to function as a question sentence generating device for use with a question-answering system, wherein said computer is connected to a question sentence generating database comprising a plurality of records for generating question sentences, each of said plurality of records having a key word and is generated from an answer having an answer sentence pattern co-occurring with the word, wherein, when executed by a processor, said computer program causes said processor to:

receive a word or a word sequence as a source for generating a question sentence; and

search and retrieve from said question sentence generating database an answer sentence pattern using a key consisting of the received word;

generate a question sentence from the retrieved answer sentence pattern and said received word, comprising:

generating question candidates by (1) transforming the retrieved answer sentence pattern into a plurality of question patterns and (2) combining each of the plurality of question patterns with the received word;

scoring each of the question candidates, wherein the score indicates a fit of each question sentence candidate as being a question having as an answer the answer from which the answer sentence pattern is generated; and

selecting as the question sentence the question sentence candidate having a best score; and; and

output the generated question sentence.

11. The computer program according to claim 10 , wherein the scoring of each of the question candidates is performed using a machine learning algorithm.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Feb 14, 2017
From: TORISAWA, KENTARO; GOTO, JUN; KLOETZER, JULIEN; KAWADA, TAKUYA
To: NATIONAL INSTITUTE OF INFORMATION AND COMMUNICATIONS TECHNOLOGY
Reel/Frame 041252/0722 →
Priority Claims (1)
JP 2014-168702 · Aug 21, 2014 · national
Continuity (1)
Related Publication 20170242915A1 · Aug 24, 2017