Efficient Text Mining Model with Conceptual Informative Relational Measure using Semantic Ontology

Abstract

The problem of text mining has been well studied and numerous approaches are analyzed towards their performance in text mining. The existing methods suffer to achieve higher performance as they consider only content of document and the term features available. Also, they measure the similarity between documents on the term features to identify the class of any document. This affects the performance of text mining and produces poor accuracy and generates higher irrelevancy. To improve the performance, a Conceptual Informative Relational Model (CIRM) is presented in this paper. Unlike previous methods, the method considers both conceptual and informative relations in measuring the similarity between the documents. The method preprocesses the text documents by eliminating the stop words, stemming and identifies list of root words or nouns. The root words extracted has been used to measure the conceptual relation and informative relation according to the taxonomy of classes and semantic meanings. Based on the value of relational measures, the method identifies the class of the document and produces result set. The proposed method improves the performance of text mining and reduces the irrelevancy.