Abstract

An chinese text classification algorithm based on graph space model

Author(s): Xiaoqiang Jia

In the field of information processing,most of the existing text classification algorithm is based on vector space model, but vector space model is not able to effectively express the document structure information, so that it is not enough to express the semantic information of documents context. In order to get more semantic information effectively, by the study of text representation of graph space model, use Common node structural equivalence and Common chain structure equivalence, analyse nodes and edges of themaximumcommon substructure graph, and judge which if is a true semantic equivalence. Next, a data structure for text classification on Graph space model was designed. On the basis of structural equivalence analysis, the distance formula of “MCS” has been improved, then an improved text similaritymetric algorithmbased on the graph space model has been proposed, experiments show that the text classification method is effective and feasible.