| 摘要 |
PROBLEM TO BE SOLVED: To achieve power saving in generation of an appropriate keyword set to be used for ranking or the like. SOLUTION: From a related word storage part, a related word corresponding to each seed word stored in a seed word storage part in which a seed word and a flag indicating the presence/absence of user support corresponding to the seed word are stored, and as for each extracted related word, a closeness score indicating the degree of closeness corresponding to the pertinent seed word to be adjusted according to a user support rate to be calculated from a flag state is calculated for each seed word, and as for each extracted related word, a relevance being the total sum of the product of the closeness score corresponding to each seed word and a coefficient to be specified according to the flag of the pertinent seed word is calculated, and a related word is extracted according to the calculated relevancy, and presented to the user as the candidate word, and as for each candidate word, data indicating the presence/absence of user support are received, and when user support is present in all the candidate words, the candidate words are registered in a dictionary, and when user support is not present even in any one candidate, those candidate words are stored in the seed word storage part as the seed words. COPYRIGHT: (C)2011,JPO&INPIT |
