基于用户兴趣群集模型的个性化元搜索研究

发布时间：2018-06-22 06:58

本文选题：元搜索引擎 + 用户兴趣群集模型　；参考：《安徽工业大学》2012年硕士论文

【摘要】：随着web技术的发展，网络资源呈现爆炸式的增长，搜索引擎已成为人们日常生活中查找资源的必备工具。而现有的搜索引擎存在无法返回用户所需的全部信息以及返回的信息大部分与用户的兴趣无关等主要问题。元搜索引擎的出现解决了独立搜索引擎覆盖率不足和查准率不高的问题。然而现有元搜索引擎建模技术的研究，往往以单个用户为基点建立用户兴趣模型，而忽略了用户共同的兴趣；注重用户查询兴趣模型，而忽略了用户对成员搜索引擎的偏好、用户收藏和用户友好等与社会化搜索相关的用户兴趣信息。本文从用户共同兴趣出发，建立了用户兴趣群集模型，设计和实现了基于用户兴趣群集模型的元搜索引擎GMS（Group Meta Search）。本文的主要研究工作如下： 1、分析了用户搜索的行为方式和特点，提出了基于用户搜索行为的兴趣度计算方法。 2、提取多个用户兴趣的主要特征，分析出多个用户搜索的兴趣类；采用Beeferman聚类算法形成不同的用户群，，并在本体模型的基础上建立用户兴趣群集模型。 3、提出基于用户兴趣群集模型的相关推荐算法。通过用户对成员搜索引擎的偏好调整各搜索引擎在元搜索引擎中的权重；结合了成员搜索引擎的推荐结果和用户兴趣群集模型推荐结果两方面改进了元搜索引擎的排序算法。 4、设计和实现了基于用户兴趣群集模型的元搜索引擎GMS（Group Meta Search），并用于推荐算法的仿真。最后，论文分别采用不同的搜索主题，对GMS元搜索引擎进行了性能测试。实验结果表明，基于用户兴趣群集模型的元搜索引擎GMS能更好地反映用户共同的兴趣特征，在个性化推荐服务方面也有更好的表现。结合用户对搜索引擎的需求趋势，还指明了下一步研究的方向。
[Abstract]:With the development of web technology and the explosive growth of network resources, search engine has become a necessary tool to find resources in people's daily life. However, the existing search engines have the main problems such as not returning all the information needed by the user, and most of the information returned is irrelevant to the interest of the user, and so on. The emergence of meta-search engine solves the problem of insufficient coverage and low precision of independent search engine. However, the existing research of meta-search engine modeling technology often builds user interest model based on a single user, but neglects the common interest of users, pays attention to user query interest model, and neglects user preference to member search engine. User collection and user-friendly and other social search related to user interest information. Based on the common interests of users, this paper establishes a cluster model of user interest, and designs and implements a meta search engine GMS (Group Meta search) based on user interest cluster model. The main research work of this paper is as follows: 1. The behavior and characteristics of user search are analyzed, and the method of calculating interest degree based on user search behavior is proposed. 2. The interest classes of multiple users are analyzed, and different user groups are formed by Beeferman clustering algorithm, and the user interest cluster model is established on the basis of ontology model. 3. A related recommendation algorithm based on user interest cluster model is proposed. Adjust the weight of each search engine in the meta search engine through the user's preference to the member search engine; Combining the recommended results of member search engines and the recommended results of user interest cluster model, the sorting algorithm of meta search engine is improved. 4. A meta-search index based on user interest cluster model is designed and implemented. GMS (Group Meta search), and used in the simulation of recommendation algorithm. Finally, the performance of GMS meta-search engine is tested with different search topics. The experimental results show that the meta-search engine GMS based on user interest cluster model can better reflect the common interest characteristics of users and has better performance in personalized recommendation service. Combined with the trend of users' demand for search engine, it also points out the direction of further research.
【学位授予单位】：安徽工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.3

【参考文献】