SIMILARITY INDEX COMPUTATION DEVICE, SIMILARITY SEARCH DEVICE, AND SIMILARITY INDEX COMPUTATION PROGRAM

摘要

Provided is a similarity index computation device comprising: a word extraction part 11 for analyzing m texts to extract n words; a vector computation part 12 for vectorizing the m texts and the n words respectively in q dimensions, thereby computing m text vectors formed from q axial components and n word vectors formed from q axial components; and an index computation part 13 for obtaining the inner products of the m text vectors and the n word vectors, thereby computing a similarity index reflecting the relation between the m texts and the n words. By calculating the inner products of the text vectors computed from the texts and the word vectors computed from the words included in the texts, it is possible to obtain a similarity evaluation value representing, as an inner product value, the extent to which a given word contributes to a given text.