The document categorization by the vector space model using lexical different area Co-occurrence

摘要

We're asked for high quality classification of documents, even if document sets have field polysemy. A noun pair with 0.15 to 0.35 or more Euclid distance, between two nouns which appear in the same sentence of a document, is called a lexical different area co-occurrence, and we propose the document categorization using this method. Using a newspaper an industry magazine, and a paper as a document set, we prove that, (1) We acquired about 90% of accuracy by the newspaper. (2) A lexical different area co-occurrence developed the 15% distance between fields than the single noun.