부산대학교 도서관

Text mining is a growing in a unique field that attempts to convey information from natural language text. It is too strength be uncertainty characterized as the development of analysing text as well as to take out the information that is constructive for particular purposes. Compared with other kind of data stored in databases, text is unstructured, shapeless as well as difficult to deal with algorithmically. Traditionally the patterns of the text are analyzed by using PTM and FTM algorithms but this algorithms produces less accuracy. To improve the performance of pattern classification, in this paper an Enhanced Support Vector Machine based Pattern classification method(ESVMPCM) is proposed. Then ESVMPCM is applied to Reuters dataset and measure the performance using different metrics as precision, recall, F-measure and accuracy.