Inference-driven multi-source semantic search

• US 8,200,656 B2
• Filed: 11/17/2009
• Issued: 06/12/2012
• Est. Priority Date: 11/17/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Brown, Eric W., Chu-Carroll, Jennifer, Fan, James J., Ferrucci, David A., Gondek, David C., Levas, Anthony T., Murdock, James William IV
• Primary Examiner(s): Leroux, Etienne
• Application Number:
  US12/620,279
• Publication Number:
  US 20110119254A1
• Time in Patent Office: 938 Days
• Field of Search: 707/2, 707/718, 707/713, 705/2
• US Class Current: 707/713
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/36 Creation of semantic tools, e.g. ontology or thesauri

Abstract

A method, system and computer program product are disclosed for searching for information using a knowledge base. In one embodiment, the method comprises receiving a query; formulizing the query, including dividing the query into a plurality of parts; for each of the parts, identifying a source, using the knowledge, that addresses that part; and combining the sources to answer the query. In one embodiment, the query includes text; the text is separated into a plurality of segments; and, for each of the segments, at least one source is identified addressing the segment. In an embodiment, a logical proof is formulated having a conclusion that is an answer to the query, and a sequence of statements that establish said conclusion; and a proof of this conclusion is generated by identifying two or more documents that assert the sequence of statements.

Claims

1. A method of searching for information using a knowledge base, comprising:

receiving a query;

formulizing the query, including dividing the query into a plurality of parts and forming a semantic representation of each of the plurality of parts;

for each of said parts, identifying one or more documents, using the knowledge base and the semantic representation of said each part, that provides relevant information for said each part in a defined manner; and

combining the information from said documents to provide an answer to the query.

2. The method according to claim 1, wherein:

the query includes text;

the dividing includes separating the text of the query into a plurality of text segments; and

the identifying includes, for each of the text segments, identifying said one or more documents, by using the knowledge base, that provides relevant information for said each text segment in said defined manner.

3. The method according to claim 1, wherein:

the dividing includes formulating a logical proof having a conclusion that is an answer to the query, and a sequence of statements that establish said conclusion; and

the identifying includes generating a proof of said conclusion by using the knowledge base to identify a set of two or more documents that assert said sequence of statements.

4. The method according to claim 3, wherein each of said sequence of statements is asserted by one of said documents.

5. The method according to claim 4, wherein no one of the documents proves said conclusion, and said proof requires all of the documents in said set of documents.

6. The method according to claim 3, wherein the generating includes generating a plurality of proofs of said conclusion, each of said proofs having an associated sequence of statements.

7. The method according to claim 6, wherein the generating a plurality of proofs includes using the knowledge base to identify, for each of said plurality of proofs, an associated set of documents that assert the sequence of statements associated with said each proof.

8. The method according to claim 7, wherein the generating a plurality of proofs further includes ordering the sets of documents according to a given one or more criteria.

9. The method according to claim 3, wherein the knowledge base includes a repository of assertions, and the method comprises the further step of building said repository from a defined group of documents.

10. The method according to claim 9, wherein:

said building includes linking each of the assertions to one of the documents, of said defined group of documents, that asserts said each assertion; and

the generating includes, for each of said statements, identifying one of the assertions in the knowledge base that asserts said statement, and using said identified assertions and said linking to identify one of the documents, of said defined group of documents, that assert said identified assertion.

11. A system for searching for information using a knowledge base, comprising one or more processor units configured for:

receiving a query;

formulizing the query, including dividing the query into a plurality of parts and forming a semantic representation of each of the plurality of parts;

for each of said parts, identifying one or more documents, using the knowledge base and the semantic representation of said each part, that provides relevant information for said each part in a defined manner; and

combining the information from said documents to provide an answer to the query.

12. The system according to claim 11, wherein:

the query includes text;

the dividing includes separating the text of the query into a plurality of text segments; and

the identifying includes, for each of the text segments, identifying said one or more documents, by using the knowledge base, that provides relevant information for said each text segment in said defined manner.

13. The system according to claim 11, wherein:

the dividing includes formulating a logical proof having a conclusion that is an answer to the query, and a sequence of statements that establish said conclusion; and

the identifying includes generating a proof of said conclusion by using the knowledge base to identify a set of two or more documents that assert said sequence of statements.

14. The system according to claim 13, wherein the generating includes:

generating a plurality of proofs of said conclusion, each of said proofs having an associated sequence of statements;

using the knowledge base to identify, for each of said plurality of proofs, an associated set of documents that assert the sequence of statements associated with said each proof; and

ordering the sets of documents according to a given on or more criteria.

15. The system according to claim 13, wherein:

the knowledge base includes a repository of assertions built from a defined group of documents;

each of the assertions is linked to one of the documents, of said defined group of documents, that asserts said each assertion; and

the generating includes, for each of said statements, identifying one of the assertions in the knowledge base that asserts said statement, and using said identified assertions and said links to identify one of the documents, of said defined group of documents, that assert said identified assertion.

16. An article of manufacture comprising:

at least one tangible computer readable medium having computer readable program code logic to execute machine instructions in one or more processing units to search for information using a knowledge base, said computer readable program code logic, when executing, performing the following;

receiving a query;

formulizing the query, including dividing the query into a plurality of parts and forming a semantic representation of each of the plurality of parts;

for each of said parts, identifying one or more documents, using the knowledge base and the semantic representation of said each part, that provides relevant information for said each part in a defined manner; and

combining the information from said documents to provide an answer to the query.

17. The article of manufacture according to claim 16, wherein:

the query includes text;

the dividing includes separating the text of the query into a plurality of text segments; and

the identifying includes, for each of the text segments, identifying said one or more documents, by using the knowledge base, that provides relevant information for said each text segment in said defined manner.

18. The article of manufacture according to claim 16, wherein:

the dividing includes formulating a logical proof having a conclusion that is an answer to the query, and a sequence of statements that establish said conclusion; and

the identifying includes generating a proof of said conclusion by using the knowledge base to identify a set of two or more documents that assert said sequence of statements.

19. The article of manufacture according to claim 18, wherein the generating includes:

generating a plurality of proofs of said conclusion, each of said proofs having an associated sequence of statements; and

using the knowledge base to identify, for each of said plurality of proofs, an associated set of documents that assert the sequence of statements associated with said each proof.

20. The article of manufacture according to claim 18, wherein:

the knowledge base includes a repository of assertions built from a defined group of documents, and each of the assertions is linked to one of the documents; and

the generating includes, for each of said statements, identifying one of the assertions in the knowledge base that asserts said statement, and using said identified assertions and said links to identify one of the documents, of said defined group of documents, that assert said identified assertion.