document search keyword presentation apparatus and method and program

摘要

PROBLEM TO BE SOLVED: To calculate recommended keyword on the basis of statistics of a sufficient number of pairs of place names and keywords without sacrificing validity of recommended keywords.;SOLUTION: Candidate representations A, positions of the candidate representations A in documents, and place name representations in the documents are extracted from a set of documents, and a plurality of extraction criteria different by quality are stepwise applied to the candidate representations A and meshes to obtain cooccurrence frequencies and appearance frequencies of the candidate representations A and the meshes in cooccurrence ranges corresponding to the extraction criteria. On the basis of pairs of meshes and candidate representations B included in a target map range inputted from a user, the cooccurrence frequency or appearance frequency is referred to in the order from the extraction criterion for the narrowest cooccurrence range or appearance range to calculate scores showing whether the number of pairs reaches a sufficient quantity or not. If the scores show that the number of pairs reaches the sufficient quantity, this cooccurrence range or the appearance range is adopted, but if the scores does not show it, the range is extended to calculate scores, and N pairs for which higher scores are calculated are outputted.;COPYRIGHT: (C)2014,JPO&INPIT