Mondou: interface with text data mining for Web search engine

Abstract

In order to submit queries to Web search engines, we have to carefully choose the suitable combination of keywords. Without rich background knowledge about keywords in Web documents, it is too difficult to find out invaluable URLs by search engines. Applying techniques of text data mining to Web resource discovery, we try to derive associative keywords by an extended association algorithm. We explain the interface of the Web resource discovery system using association rules, which are derived from the cluster of Japanese HTML pages in the text database. We discuss an evaluation of the Mondou system and Java applet order to visualize search results with multidimensional measurements.