A Study of Document Search using Correspondence Analysis

Abstract

The opportunity that we want to search target document(s) among a large amount of electronic documents such as WEB information and documents inputted by document scanner will be more increasing in near future. We focus on the approach that applies the distribution pattern of the keyword frequency in documents to the search of the documents. In this paper, we propose the method that uses the correspondence analysis for the search of documents based on the similarity of the keyword pattern among them. We apply the proposed method to the data of patents of Nagoya University, and report the result in terms of the accuracy for the document search. The experimental result shows that we can find the related patents to the inputted keyword by a user.