关键词: 搜索引擎日志, 点击模型, 文档相关性特征, 浏览时间

Abstract:

As the clicks of users are affected by many factors such as the browsing behaviors of users and the features of adjacent documents,they cannot be directly interpreted as document relevance during the analysis of users' click log. In order to obtain correct document relevance features,a click model based on browsing time is proposed. In this model,the document relevance is assumed to affect the browsing time for the document,which means a user may spend more browsing time if the document is more likely to be relevant. Therefore,the document relevance can be estimated according to the browsing time. Then,an exponential distribution is determined to describe users' browsing time,and some experiments are carried out to verify the effectiveness of the proposed model. The results indicate that the model is superior to the existing click models because it can estimate document relevance and future clicks more accurately.

Key words: search engine log, click model, document relevance feature, browsing time