Efficient multi objective global optimization approach for feature selection in text classification

Abstract

Text document management has become an essential methodology due to the rapid growth in document digitization through World Wide Web (WWW). The Biomedical Information Retrieval (BIR) system is used for the effective retrieval of biomedical information from published data. The BIR system systematically compares large data sets with relevant biological data sets from the available data. Text classification plays a vital role in text mining and is classified as topic-based text mining and genre-based mining text. The objective of medi-facts retrieval can be achieved by efficient text classification approaches. Automatic text classification classifies text documents based on the predefined texts. High dimensionality in a feature space is a major issue in text categorization. The identification of informative features in a set of documents reduces the dimensionality of feature space. The selection of informative feature in a high dimension is an NP hard problem and hence meta-heuristic approaches can be used. In 2002, newline newline newline