Predicting Click-Through Rate Using Category Features Extracted from Advertisement

摘要：

　　The click-through rate (CTR) of online advertisement is a crucial indicator for search engine providers and advertisers.A method for CTR prediction based on category features extracted from advertisement is proposed in this paper.This method defines the trigger effect of usersquery on advertisement,extracts multiple labels of one advertisement through indirect clustering and predicts the CTR by inputting category features and original factorized features with the Factorization Machines of prediction model.The experiment result manifests that category features of advertisement significantly enhance the accuracy of prediction; the indirect clustering used during feature extraction labels multiple categories of advertisement in such a way that the dimension of eigenvector is effectively reduced and time cost for clustering is dramatically decreased.

关键词： click-through rate feature extraction indirect clustering factorization machines

作者: Tang Liu Chuang Zhang Tao Xia

作者单位: Pattern Recognition and Intelligent System Laboratory Beijing University of Posts and Telecommunications Beijing 100876, China

会议类型: 国际会议

会议名称: 2012 IEEE 14th International Conference on Communication Technology(2012年第十四届通信技术国际会议(ICCT 2012))

会议地点: 成都

会议语种:英文

页码: 1589-1593

在线出版日期: 2012-11-09（万方平台首次上网日期，不代表论文的发表时间）