Meaningful Inner Link Objects for Automatic Text Categorization

This paper presents a novel approach for automatic text categorization. The mainstream of the research on rule-based classifier regards document as a container of term, and generates rules by using the term distribution in documents. General speaking, there must be existed some kind of semantic relevance between term and paragraph in a document. We call it Meaningful Inner Link Objects-MILO which must be varied with different semantics of a document itself. While this paper concentrates on using these MILOs that associate with semantic relevance for text categorization, hence we focus on two problems: (1) finding the best MILOs which associate with semantic relevance; and (2) using these specific MILOs to build a classifier for text categorization. From the experiment results, our proposed classification approach base on MILO has a better accuracy while other state of the art technique without considering the relevance between term and paragraph.