SIMILAR DOCUMENT SET EXTRACTION DEVICE, SIMILAR DOCUMENT SET EXTRACTION METHOD, SIMILAR DOCUMENT SET EXTRACTION PROGRAM AND STORAGE MEDIUM

摘要

PROBLEM TO BE SOLVED: To improve the similarity accuracy of an extracted similar document set.;SOLUTION: The similar document set extraction device 1 for extracting a similar document set from documents accumulated in a document database 100 comprises an input means 10 inputting a numeric value showing the number of similar document sets to be extracted; a document set extraction processing part 52 executing extraction operation of similar document set by the frequency of the input numeric value based on an evaluation function; and an output means 20 outputting the extracted similar document sets. The evaluation function is obtained by summing up the difference between the similarity of a word vector characterizing a document to a representative vector of a document set containing this document and the similarity of the word vector characterizing the document to a representative vector of an extracted document set containing this document over each document contained in the document set. A similar document set extraction processing part 53 determines a value of evaluation function for each document set, and extracts a document set in which the value of evaluation function is maximum.;COPYRIGHT: (C)2007,JPO&INPIT