基于时间敏感的个性化查询词补全算法研究

发布时间：2018-05-27 22:34

本文选题：个性化查询词补全 + 时间敏感分析　；参考：《北京林业大学》2015年硕士论文

【摘要】：当今互联网信息纷繁复杂,搜索引擎为互联网用户提供了良好的信息搜索体验。而搜索引擎的查询词补全技术又进一步提高了搜索引擎质量,为用户提供了更优质的搜索服务。目前查询词补全技术广泛应用于各搜索引擎,但大多依赖群体智慧策略。针对用户检索需求随时间变化而不同这一问题,时间敏感查询词自动补全技术越来越成为研究热点。时间敏感查询词补全在生成查询词补全候选列表时拟合多种时间因素,呈现出与传统查询词补全不同的特点。本文针对查询词自动补全技术非个性化及时间敏感化的问题,提出了一个方法“个性化时间敏感查询词补全”来推荐时间敏感的个性化查询。首先,将用户所有桌面操作行为数据组织建立索引,并存储于用户个人计算机中；其次,用贝叶斯公式描述用户需求和时间的关联,并应用时间序列预测模型计算时间敏感值；最后,根据时间敏感权值对所有补全候选词进行排序。通过最终的在真实数据集上的实验结果对比证明,论文模型可以用于加强信息检索,特别是查询词补全推荐相关领域算法。
[Abstract]:Nowadays, Internet information is complicated, search engine provides good information search experience for Internet users. Search engine query word complement technology further improves the search engine quality and provides users with better search services. At present, query word complement technology is widely used in various search engines, but mostly rely on group intelligence strategy. In order to solve the problem that users' retrieval requirements vary with time, the automatic completion of time-sensitive query terms has become a hot research topic. Time-sensitive query word complement is different from traditional query word complement because of fitting various time factors when generating query word complement candidate list. Aiming at the problem of non-personalization and time sensitivity of automatic completion of query words, this paper proposes a method of "personalized time-sensitive query word complement" to recommend time-sensitive personalized query. First of all, the data of user's desktop operation behavior are organized and indexed, and stored in the user's personal computer. Secondly, Bayesian formula is used to describe the relationship between user's demand and time, and time sensitive value is calculated by using time series prediction model. Finally, all the complement candidate words are sorted according to the time sensitive weights. The experimental results on the real data set show that the model can be used to enhance information retrieval, especially the algorithm of complete recommendation of query words.
【学位授予单位】：北京林业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.3

【相似文献】