Removing ambiguity when analyzing a sentence with a word having multiple meanings

• US 8,301,435 B2
• Filed: 02/09/2007
• Issued: 10/30/2012
• Est. Priority Date: 02/27/2006
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Sadamasa, Kunihiko, Ando, Shinichi, Doi, Shinichi
• Primary Examiner(s): He, Jialong
• Application Number:
  US12/280,839
• Publication Number:
  US 20090018821A1
• Time in Patent Office: 2,090 Days
• Field of Search: 704 1- 10
• US Class Current: 704/2
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A language processing device includes first analysis unit 21 that subjects a natural language sentence containing a polysemic word and other words to a predetermined analysis and outputs a plurality of analysis results for the natural language sentence according to a plurality of meanings of the polysemic word, second analysis unit 23 that performs a particular analysis on the analysis results outputted from first analysis unit 21, and employs one of the analysis results, and generation unit 244 that generates a deletion rule for deleting one or more unnecessary analysis results of the first analysis unit 21 which has been deleted from the analysis results outputted from first analysis unit 21 but employed by second analysis unit 23, according to the analysis results outputted from the first analysis unit 21 and the employment result of second analysis unit 23.

Claims

1. A language processing device comprising:

a first analysis unit that performs a predetermined analysis on a natural language sentence including a polysemic word and other words to output a plurality of analysis results for the natural language sentence in accordance with a plurality of meanings possessed by the polysemic word;

a second analysis unit that performs a particular analysis on the plurality of analysis results outputted by said first analysis unit to adopt one analysis result from the plurality of analysis results;

a storage unit that stores determination information, for each combination of one meaning of said polysemic word with information related to the other words, the determination information indicating whether or not each combination is adopted;

an adoption information generation unit that generates adoption information for each combination of the meaning of the polysemic word within each analysis result outputted by said first analysis unit with the information related to the other words, the adoption information indicating whether or not each combination is adopted by said second analysis unit; and

a rule generation unit that, when a first said combination that is indicated to be “

not adopted”

in the adoption information corresponds to the determination information that indicates the first combination is “

adopted”

, the rule generation unit changes the adoption information for the first combination to “

adopted” and

subsequently generates a deletion rule for deleting one or more unnecessary analysis results from the plurality of analysis results, the deletion rule being based on the updated adoption information.

2. The language processing device according to claim 1, wherein each said combination of the meaning of the polysemic word with information related to the other words includes one of the following:

a surface string, a word class, a conjugation, and a combination thereof of the polysemic word with words therearound.

3. A language processing method performed by a language processing device embodied in a computer and including a storage unit which stores determination information, for each combination of one meaning of a polysemic word with information related to other words, that indicates whether or not the combination is adopted, said method comprising:

performing, in the computer, a predetermined analysis on a natural language sentence including a polysemic word and other words to output a plurality of analysis results for the natural language sentence in accordance with a plurality of meanings possessed by the polysemic word;

performing, in the computer, a particular analysis on the plurality of analysis results to adopt one analysis result from the plurality of analysis results;

generating, in the computer, adoption information, for each combination of the meaning of the polysemic word within each analysis result outputted by the step of performing the predetermined analysis with the information related to the other words, the adoption information indicating whether or not each combination is adopted by the step of performing the particular analysis; and

when a first said combination that is indicated to be “

not adopted”

in the adoption information corresponds to the determination information that indicates the first combination is “

adopted”

, changing the adoption information for the first combination to “

adopted” and

subsequently generating a deletion rule for deleting one or more unnecessary analysis results from the plurality of analysis results, the deletion rule being based on the updated adoption information.

4. A non-transient computer readable recording medium in which a language processing program is embedded, said program causing a computer to execute language processing, said computer being connected to a storage unit which stores determination information, for each combination of a meaning of a polysemic word with information related to other words, that indicates whether or not the combination is adopted, said language processing comprising:

first analysis processing for performing a predetermined analysis on a natural language sentence including a polysemic word and other words to output a plurality of analysis results for the natural language sentence in accordance with a plurality of meanings possessed by the polysemic word;

second analysis processing for performing a particular analysis on the plurality of analysis results to adopt one analysis result from the plurality of analysis results;

adoption information generation processing for generating adoption information for each combination of the meaning of the polysemic word within each analysis result outputted by said first analysis processing step with the information related to the other words, the adoption information indicating whether or not the combination is adopted by said second analysis processing step; and

rule generation processing for, when a first said combination is indicated to be “

not adopted”

in the adoption information corresponds to the determination information that indicates the first combination is “

adopted”

, changing the adoption information for the first combination to “

adopted” and

subsequently generating a deletion rule for deleting one or more unnecessary analysis results from the plurality of analysis results, the deletion rule being based on the updated adoption information.

5. A language processing device comprising:

first analysis means for performing a predetermined analysis on a natural language sentence including a polysemic word and other words to output a plurality of analysis results for the natural language sentence in accordance with a plurality of meanings possessed by the polysemic word;

second analysis means for performing a particular analysis on the plurality of analysis results outputted by said first analysis means to adopt one analysis result from the plurality of analysis results;

storage means for storing determination information for each combination of one meaning of said polysemic word with information related to the other words, the determination information indicating whether or not the combination is adopted;

adoption information generation means for generating adoption information for each combination of the meaning of the polysemic word within each analysis result outputted by said first analysis means with the information related to the other words, the adoption information indicating whether or not the combination is adopted by said second analysis means; and

rule generation means for, when a first said combination is indicated to be “

not adopted”

in the adoption information corresponds to the determination information that indicates that the first combination is “

adopted”

, changing the adoption information for the first combination to “

adopted” and

subsequently generating a deletion rule for deleting one or more unnecessary analysis results from the plurality of analysis results, the deletion rule being based on the updated adoption information.

6. The language processing device of claim 5, further comprising knowledge base storage means for storing the deletion rule;

and

ambiguity deletion means for deleting one or more unnecessary analysis results from the plurality of analysis results outputted by said first analysis means based on the deletion rule stored in said knowledge base storage means.