System and method of inferring synonyms using ensemble learning techniques

• US 10,832,146 B2
• Filed: 01/19/2016
• Issued: 11/10/2020
• Est. Priority Date: 01/19/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Kaufmann, Joseph M., Krishnamurthy, Lakshminarayanan, Mendes, Pablo N.
• Primary Examiner(s): Chang, Li Wu
• Application Number:
  US15/000,582
• Publication Number:
  US 20170206453A1
• Time in Patent Office: 1,757 Days
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

Embodiments are directed to a method of utilizing an ensemble of distributional semantics systems in conjunction with a domain term extractor for generating domain-specific synonyms. The method allows for extraction of high-quality, domain-specific synonyms that can be used in an information handling system, such as a question-answer system or in an information retrieval (IR) system, capable of processing natural language. According to embodiments, the domain term extractor identifies the words for which synonyms are sought, and the ensemble of distributional semantics systems determines the synonyms.

Claims

1. A computer implemented method in a data processing system comprising a processor and a memory comprising instructions which are executed by the processor to cause the processor to implement a method for utilizing information sources to identify domain-specific synonyms for an information handling system capable of processing natural language, the method comprising:

receiving, by the processor, a domain-specific information source;

identifying, by the processor, domain-specific terms to form a candidate synonym list, by comparing each term of the domain-specific information source to an open domain source;

applying, by the processor, at least two ensemble machine learning techniques to train a synonym finder, wherein the synonym finder is a distributional semantics ensemble comprising at least two distributional semantics systems, wherein each distributional semantics system is trained separately, wherein one distributional semantics system is DiSSect, and another distributional semantics system is Glimpse; and

utilizing, by the processor, the synonym finder to find synonyms of terms in the candidate synonym list.

2. The method of claim 1, wherein:

the at least two ensemble machine learning techniques utilize a first context representation, a second context representation different from the first context representation, a first transformational technique, and a second transformational technique different from the first transformational technique.

3. The method of claim 1, further comprising:

utilizing, by the processor, the synonyms for one or more of a question-answer system and an information retrieval system.

4. The method of claim 1, wherein each of the at least two distributional semantics systems is individually trained.

5. The method of claim 1, wherein the applying of the at least two ensemble machine learning techniques to train the synonym finder comprises applying crowdsourcing techniques to obtain training data.

6. A system for identifying domain-specific synonyms for an information handling system capable of processing natural language, comprising:

a processor configured to;

receive a domain-specific information source;

identify domain-specific terms to form a candidate synonym list, by comparing each term of the domain-specific information source to an open domain source;

apply at least two ensemble machine learning techniques to train a synonym finder, wherein the synonym finder is a distributional semantics ensemble comprising at least two distributional semantics systems, wherein each distributional semantics system is trained separately, wherein one distributional semantics system is DiSSect, and another distributional semantics system is Glimpse; and

utilize the synonym finder to find synonyms of terms in the candidate synonym list.

7. The system of claim 6, wherein the processor is further configured to:

utilize the synonyms for one or more of a question-answer system and an information retrieval system.

8. The system of claim 6, wherein each of the at least two distributional semantics systems is individually trained.

9. The system of claim 6, wherein the applying, by the processor, of the at least two ensemble machine learning techniques to train the synonym finder comprises applying crowdsourcing techniques to obtain training data.

14. The system of claim 6, wherein:

the at least two ensemble machine learning techniques utilize a first context representation, a second context representation different from the first context representation, a first transformational technique, and a second transformational technique different from the first transformational technique.

10. A computer program product for identifying domain-specific synonyms for an information handling system capable of processing natural language, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

receive, by the processor, a domain-specific information source;

identify, by the processor, domain-specific terms to form a candidate synonym list, by comparing each term of the domain-specific information source to an open domain source;

apply, by the processor, at least two ensemble machine learning techniques to train a synonym finder, wherein the synonym finder is a distributional semantics ensemble comprising at least two distributional semantics systems, wherein each distributional semantics system is trained separately, wherein one distributional semantics system is DiSSect, and another distributional semantics system is Glimpse; and

utilize, by the processor, the synonym finder to find synonyms of terms in the candidate synonym list.

11. The computer program product of claim 10, wherein:

the at least two ensemble machine learning techniques utilize a first context representation, a second context representation different from the first context representation, a first transformational technique, and a second transformational technique different from the first transformational technique.

12. The computer program product of claim 10, wherein the program instructions executable by the processor further cause the processor to:

utilize, by the processor, the synonyms for one or more of a question-answer system and an information retrieval system.

13. The computer program product of claim 10, wherein each of the at least two distributional semantics systems is individually trained.