基于用户日志的相关搜索模型研究

Related search on model based on user logs

搜索技术的组成部分发展至今已经呈现多样化,在不同的使用环境它们各有特色,但都致力于为使用者提供更优质的使用体验。现代搜索技术希望从使用者的自然语言出发,为使用者发掘到最理想的目标答案。相关搜索是查询推荐技术在搜索引擎中最常见和成功的实现,传统相关搜索基于日志建立的模型是为了应付基于内容分析或初次检索模型的局限性。文中重新划分粒度,更加精确计算用户日志数据的相关性,为使用者提供更好的相关推荐。实验表明,文章结果在应用中有了较好的提高。

The search technology has been an integral part of diversified, it has different characteristics in order to apply in different scenarios. But all committed to provide users with better experience, and hope to explore the best way for users to target answers with the user＇s natural language. Related search is the specific performance of using the recommended techniques. Some of the traditional logs mining methods want to cope with the limitations of based on content analysis or initial retrieval. The paper established a new model, and the re-division of granularity, more accurate calculation of the relevant user logs data. The experimental results show, compared with the model has been used the results have been significantly improved.