关键词: Web文档, 聚类, 概念格, 约简

Abstract: Web text clustering are mostly based on space vector text express model，the semantics relation of the terms in the text is not considered in this method and the dimension of the terms is very high，which results in the losing of text semantics and the increase of time complexity.The text is considered as object in this paper，and the term of text is abstract as the corresponding attribute.Therefore，a formal context is formed based on text.To express text and measure the similarity the authors extract the concept from formal context.Thus，the dimension of term is reduced，and the complexity of computation is decreased too.Theoretical analysis shows that the method of clustering is effective.

Key words: Web document, clustering, concept lattice, reduce