Document retrieval device

Abstract

Parts of documents are retrieved using the entire context of selected documents. Classification unit designation section performs the designation of a classification unit. A logical structure analysis section analyzes the logical structure of the documents read-in from a document storing section where the documents are stored. A fundamental vector generation section partitions the logical structure of the documents by means of the classification unit, extracts keywords, and generates fundamental vectors. A heading vector generation section extracts key words from the headings of the structural elements that are arranged in higher level of structure than the structural element of the classification unit that was the target of fundamental vector generation, and generates heading vectors. A vector synthesis section synthesizes fundamental vectors and heading vectors, and generates composite vectors. Composite vector maintenance section attaches the corresponding composite vectors to structural elements of the classification unit that were the target of composite vector generation and maintains the attached objects. A classification section classifies the structural elements of the documents of the classification unit based on the degree of similarity of the generated composite vectors. A display section displays the results of classification.