Feedback learning classifier based on TFIDF

摘要：

Traditional feature weighting algorithm TFIDF doest take some other factors which impact the feature weight into consideration, so this paper discusses the factors in details and proposes a new feature weighting algorithm called NTFIDF combined with these factors and TFIDF. Experiment on the KNN classifier shows that NTFIDF is better than TFIDF in text classification. KNN classifier is based on training, which means once the sample size is fixed, the classification ability is also fixed, so this article proposes a feedback learning method which enable the classifier to keep learning to enhance the classification capability of its own. We design and implement such a feedback learning classifier.

关键词： TFIDF NTFIDF text classification KNN feedback learning

作者: Hao Jiang Wenqiang Li Anjian Pu

作者单位: School of Computer Science & Engineering Southeast University Nanjing ,China

会议类型: 国际会议

会议名称: The 13th IEEE Joint International Computer Science and Information Technology Conference(2011年第13届IEEE联合国际计算机科学与信息技术会议 JICSIT 2011)

会议地点: 重庆

会议语种:英文

页码: 290-293

在线出版日期: 2011-08-20（万方平台首次上网日期，不代表论文的发表时间）