A Kind of Text Classification Design on the Basis of Natural Language Processing

作者：Zhijuan Deng, Shaojun Zhong · 发表于：International Journal of Advancements in Computing Technology · 年份：2013 · DOI：10.4156/ijact.vol5.issue1.74 · 被引用次数：6 · 研究领域：Advanced Computational Techniques and Applications

Some researches on the text classification and its related technologies have been done. A kind of text classifier was designed. In order to eliminate the ambiguity of ambiguous fields which is of triple-length overlap type and to process the stop words in the process of pretreatment, the maximal matching method has been improved. At the same time, combining the TFIDF weight of feature term on the basis of the feature selection method of KL divergence, and the selected feature term can express the content of text more accurately. The classifier has been verified by Bayesian algorithm, simple vector distance classification and KNN algorithm, and the algorithm with the best classification effect has been found out.