摘要 |
PROBLEM TO BE SOLVED: To provide a document analysis device and a program, in which a word causing rise or fall in order is extracted on the basis of order relations among categories.SOLUTION: A document analysis device according to the embodiment includes: a document storage part for storing document data; a category storage part for storing a plurality of categories with order which classifies document data and a hierarchical structure of the categories; and a factorial word extraction part for extracting, from a group of words included in a category which is one of a plurality of categories, a word whose frequency of appearance in the category is greater than a frequency of appearance of the word in another category belonging to the same layer to which the category belongs and whose frequency of appearance in the other category decreases further as the other category is deviated further, in terms of order, from the category. |