Knowledge-rich automatic term disambiguation

• US 9,633,009 B2
• Filed: 08/01/2013
• Issued: 04/25/2017
• Est. Priority Date: 08/01/2013
• Status: Expired due to Fees

• Current Assignee: AirBNB Incorporated
• Original Assignee: International Business Machines Corporation
• Inventors: Baldwin, Tyler Shore, Alexe, Bogdan, Li, Yunyao, Stanoi, Ioana Roxana, Vaithyanathan, Shivakumar
• Primary Examiner(s): Spooner, Lamont
• Assistant Examiner(s): OGUNBIYI, OLUWADAMILOL M
• Application Number:
  US13/956,663
• Publication Number:
  US 20150039290A1
• Time in Patent Office: 1,363 Days
• Field of Search: 704 9
• CPC Class Codes: G06F 40/247 Thesauruses; Synonyms
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates

Abstract

Embodiments of the invention relate to ambiguity detection. In one embodiment, an object and a topical domain associated with the object are obtained. In this embodiment, the object includes at least one term. At least one of a plurality of information sources is analyzed based on the at least one term and the topical domain. A determination is made that object is one of ambiguous and unambiguous based on analyzing at least one of the plurality of information sources.

Claims

1. A method by an information processing system, the method comprising:

electronically communicating with at least one external information processing system;

obtaining, based on electronically communicating with the at least one external information processing system, an object and a topical domain associated with the object from the at least one external information processing system, wherein the object comprises at least one term; and

for at least one information source in a plurality of information sources coupled together in a pipelined sequence, electronically processing the object as follows

analyzing the information source based on the at least one term and the topical domain;

determining, based on the analysing, if the object is one of ambiguous and unambiguous;

in response to determining that the object is ambiguous, further determining that all instances of the object are ambiguous based on the object being ambiguous, and halting the analysis of any subsequent information source in the plurality of information sources, and outputting an indication to a display coupled to an information processing system that the object is ambiguous; and

in response to determining that the object fails to be ambiguous, determining if a subsequent information source exists in the plurality of information sources;

in response to determining that a subsequent information source exists, passing the processing of the object to the subsequent information source; and

in response to determining that a subsequent information source fails to exist, outputting an indication to the display that all instances of the object are unambiguous.

2. The method of claim 1, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of n-grams generated from at least one text corpus;

generating, based on the analyzing, a count of a number of times the at least one term appears in the text corpus;

determining if the count is greater than a threshold,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing the indication that the object is ambiguous in response to determining that the count is greater than the threshold, and

storing the indication that the object is unambiguous in response to determining that the count is less than the threshold.

3. The method of claim 1, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of ontologies; and

determining, based on the analyzing the set of ontologies, if the object comprises multiple senses within at least one of the set of ontologies,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing the indication that the object is ambiguous in response to determining that the object comprises multiple senses within the at least one of the set of ontologies, and

storing the indication that the object is unambiguous in response to determining that the object fails to comprise multiple senses within the at least one of the set of ontologies.

4. The method of claim 1, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of document clusters, wherein each document comprises a set of words associated with the at least one term, and wherein each of the set of words is associated with a weight;

determining, based on analyzing the set of document clusters, if at least one of the domain and one or more synonyms of the domain appear in N highest weighted words in at least one of the set of document clusters,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing an indication that the object is ambiguous in response to determining that at least one of the domain and one or more synonyms of the domain fails to appears in the N highest weighted words in at least one of the set of document clusters, and

storing the indication that the object is unambiguous in response to determining that at least one of the domain and one or more synonyms of the domain appears in the N highest weighted words in at least one of the set of document clusters.

5. The method of claim 1, wherein the plurality of information sources comprise at least one of:

a set of text corpora;

a set of ontologies; and

a set of clustered documents.

6. An information processing system comprising:

a memory;

a processor communicatively coupled to the memory; and

an ambiguity detector communicatively coupled to the memory and the processor, wherein the ambiguity detector is configured to perform a method comprising;

electronically communicating with at least one external information processing system;

obtaining, based on electronically communicating with the at least one external information processing system, an object and a topical domain associated with the object from the at least one external information processing system, wherein the object comprises at least one term; and

for at least one information source in a plurality of information sources coupled together in a pipelined sequence electronically processing the object as follows

analyzing the information source based on the at least one term and the topical domain;

determining, based on the analyzing, if the object is one of ambiguous and unambiguous;

in response to determining that the object is ambiguous, further determining that all instances of the object are ambiguous based on the object being ambiguous, and halting the analysis of any subsequent information source in the plurality of information sources, and outputting an indication to a display coupled to an information processing system that the object is ambiguous; and

in response to determining that the object fails to be ambiguous, determining if a subsequent information source exists in the plurality of information sources;

in response to determining that a subsequent information source exists, passing the processing of the object to the subsequent information source; and

in response to determining that a subsequent information source fails to exist, outputting an indication to the display that all instances of the object are unambiguous.

7. The information processing system of claim 6, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of n-grams generated from at least one text corpus;

generating, based on the analyzing, a count of a number of times the at least one term appears in the text corpus; and

determining if the count is greater than a threshold;

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing the indication that the object is ambiguous in response to determining that the count is greater than the threshold, and

storing the indication that the object is unambiguous in response to determining that the count is less than the threshold.

8. The information processing system of claim 6, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of ontologies; and

determining, based on the analyzing the set of ontologies, if the object comprises multiple senses within at least one of the set of ontologies,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing the indication that the object is ambiguous in response to determining that the object comprises multiple senses within the at least one of the set of ontologies, and

storing the indication that the object is unambiguous in response to determining that the object fails to comprise multiple senses within the at least one of the set of ontologies.

9. The information processing system of claim 6, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of document clusters, wherein each document comprises a set of words associated with the at least one term, and wherein each of the set of words is associated with a weight; and

determining, based on analyzing the set of document clusters, if at least one of the domain and one or more synonyms of the domain appear in N highest weighted words in at least one of the set of document clusters,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing an indication that the object is ambiguous in response to determining that at least one of the domain and one or more synonyms of the domain fails to appear in the N highest weighted words in at least one of the set of document clusters, and

storing the indication that the object is unambiguous in response to determining that at least one of the domain and one or more synonyms of the domain appears in the N highest weighted words in at least one of the set of document clusters.

10. A computer program product comprising:

a non-transitory computer readable storage medium having computer readable program code embodied therewith, the computer readable program code configured to program a processing circuit to perform a method comprising;

electronically communicating with at least one external information processing system;

obtaining, based on electronically communicating with the at least one external information processing system, an object and a topical domain associated with the object from the at least one external information processing system, wherein the object comprises at least one term; and

for at least one information source in a plurality of information sources coupled together in a pipelined sequence electronically processing the object as follows

analyzing the information source based on the at least one term and the topical domain;

determining, based on the analyzing, if the object is one of ambiguous and unambiguous;

in response to determining that the object is ambiguous, further determining that all instances of the object are ambiguous based on the object being ambiguous, and halting the analysis of any subsequent information source in the plurality of information sources, and outputting an indication to a display coupled to an information processing system that the object is ambiguous; and

in response to determining that the object fails to be ambiguous, determining if a subsequent information source exists in the plurality of information sources;

in response to determining that a subsequent information source exists, passing the processing of the object to the subsequent information source; and

in response to determining that a subsequent information source fails to exist, outputting an indication to the display that all instances of the object are unambiguous.

11. The computer program product of claim 10, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of n-grams generated from at least one text corpus;

generating, based on the analyzing, a count of a number of times the at least one term appears in the text corpus; and

determining if the count is greater than a threshold,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing the indication that the object is ambiguous in response to determining that the count is greater than the threshold, and

storing the indication that the object is unambiguous in response to determining that the count is less than the threshold.

12. The computer program product of claim 10, wherein analyzing the information source comprises:

analyzing, based on the at least one term and the topical domain, a set of ontologies; and

determining, based on the analyzing the set of ontologies, if the object comprises multiple senses within at least one of the set of ontologies,

wherein determining if the object is one of ambiguous and unambiguous further comprises

storing the indication that the object is ambiguous in response to determining that the object comprises multiple senses within the at least one of the set of ontologies, and

storing the indication that the object is unambiguous in response to determining that the object fails to comprise multiple senses within the at least one of the set of ontologies.

13. The computer program product of claim 10, wherein analyzing the at least one information source comprises:

analyzing, based on the at least one term and the topical domain, a set of document clusters, wherein each document comprises a set of words associated with the at least one term, and wherein each of the set of words is associated with a weight; and

determining, based on analyzing the set of document clusters, if at least one of the domain and one or more synonyms of the domain appear in N highest weighted words in at least one of the set of document clusters,

wherein determining, if the object is one of ambiguous and unambiguous further comprises

storing an indication that the object is ambiguous in response to determining that at least one of the domain and one or more synonyms of the domain fails to appear in the N highest weighted words in at least one of the set of document clusters, and

storing the indication that the object is unambiguous in response to determining that at least one of the domain and one or more synonyms of the domain appears in the N highest weighted words in at least one of the set of document clusters.

14. The computer program product of claim 10, wherein the plurality of information sources comprise at least one of:

a set of text corpora;

a set of ontologies; and

a set of clustered documents.