SYSTEM, METHOD, AND PROGRAM FOR COLLECTING LEARNING DATA

PROBLEM TO BE SOLVED: To create an associative concept dictionary based on high-precision associative words suitable for disambiguation of polysemic words, and make learning data collectable according to parameters, such as the number of cases and quality policies. SOLUTION: An associative concept d...

Full description

Saved in:
Bibliographic Details
Main Authors UCHIYAMA MASASHI, TSUTSUMIDA KYOTA, NAKATSUJI MAKOTO, TANAKA AKIMICHI
Format Patent
LanguageEnglish
Published 25.08.2011
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:PROBLEM TO BE SOLVED: To create an associative concept dictionary based on high-precision associative words suitable for disambiguation of polysemic words, and make learning data collectable according to parameters, such as the number of cases and quality policies. SOLUTION: An associative concept dictionary 10 collects associative words which a subject is reminded from combination of a classification label word and an evaluation word showing a semantic category to which the evaluation word is classified, and holds the collected associative word combining the evaluation word and the classification label word. An associative word index creation means 12 extracts a sentence containing the evaluation word and classification label word of the associative concept dictionary from a corpus 11 which accumulates electronic document groups, creates and saves in a DB 13, the associative word index which shows the existence of associative word in extracted sentence for each pair of the evaluation word and the classification label word. A learning data collecting means 14 collects the sentences which contain the evaluation words, classification label words, and associative words based on the index from the corpus 11, and saves them in a DB 15. COPYRIGHT: (C)2011,JPO&INPIT
Bibliography:Application Number: JP20100023593