Constructing a PU Text Classifier with Incremental Characteristic

摘要：

　　Based on Focused Crawling,the paper designs and implements a PU text classification model with some incremental characteristic.For the case of negative set in the training samples which are not clear-cut,it first obtains a credible negative set by improving 1-DNF algorithm and then iterate trains the classifier,lastly obtains the final classifier for the theme of crawling text classification.The model learns some of the positive set and negative set in each training loop,then enters into the next training.It acquires a good self-adaptability,and reaches a good precision in the context of declining training samples.

关键词： PU Classification 1-DNF Focused Crawling, Incremental Characteristic

作者: ShiYong Ling JinHong Gong

作者单位: Center of Modern Education and Technology East China Jiaotong University, Nanchang,China School of Electrical Engineer, East China Jiaotong University, Nanchang,China

会议类型: 国际会议

会议名称: 2012 2nd international Conference on Materials Science and Information Technology(2012第二届材料科学与信息技术国际会议)(MSIT2012)

会议地点: 西安

会议语种:英文

页码: 1318-1323

在线出版日期: 2012-08-24（万方平台首次上网日期，不代表论文的发表时间）