基于Multi-Agent的移动个性化搜索关键技术研究

发布时间：2018-02-03 05:07

本文关键词： Multi-Agent 个性化检索移动搜索兴趣模型上下文　出处：《湖南大学》2013年硕士论文　论文类型：学位论文

【摘要】：随着计算机网络与通信技术的快速发展，移动搜索以其便利性、准确性、个性化等突出优点备受关注。如何利用移动设备从海量信息中检索出有用的信息已成为一个非常重要的研究课题。由于移动用户查询用词与文档关键词匹配程度较低，导致查询的结果并不理想，提供的内容也没有针对性，难以满足用户的需求。因此，研究如何提高移动用户的个性化检索精度就具有了重要的理论意义和实用价值。然而要向移动用户提供比较精确的个性化信息服务，关键在于了解用户兴趣，并准确的建立用户兴趣模型。首先，本文分析个性化信息服务与Agent相关理论，然后，通过用户Agent作为建模工具，以用户为中心，基于领域特征，借助基于主题偏好向量的网页差异与检索名词的差异化策略等有效方法挖掘出用户偏好兴趣模型。然后，针对移动用户的特殊性，给出基于用户上下文信息扩展的移动检索算法，利用移动用户周围的上下文信息来扩展用户查询，该算法基于位置服务LBS获取用户周围信息，计算实体与查询之间相关性，，最后选取相关性高的词语来扩展用户查询，完善查询请求并提高了查询结果的查准率。接下来，研究个性化排序算法，加入结果反馈机制，结合模型给出针对移动用户的个性化搜索算法，优化检索结果的排序，能针对不同用户的检索输入给出个性化的领域检索服务输出，把用户感兴趣的网页置于前面，能有效的提高搜索的精确度，并根据相关反馈机制更新用户兴趣模型。最后，利用智能Agent其自治性、学习性及通信协调等特性设计移动个性化搜索引擎原型，将庞大而又复杂的系统分割成多个独立自治的智能体模块，并开发出核心模块进行仿真模拟实验。实验结果表明，加入上下文信息扩展算法后在检索结果的精度上比传统检索结果要好，并且用户对改进后的系统的满意度要比之前要提高10%-25%。
[Abstract]:With the rapid development of computer network and communication technology , mobile search has attracted much attention for its convenience , accuracy , personalization and so on . How to use mobile equipment to retrieve useful information from mass information has become a very important research topic . Firstly , this paper analyzes the theory of personalized information service and agent , then uses user agent as the modeling tool , user - centered , domain - based feature , and finds out the user preference interest model by using the difference of web pages based on the theme preference vector and the differentiation strategy of the search term . then , aiming at the particularity of mobile users , a mobile search algorithm based on user context information extension is presented , and the user query is extended by utilizing the context information around the mobile user , the algorithm is based on the location service LBS to acquire the information around the user , calculates the correlation between the entity and the query , and finally selects the words with high correlation to expand the user query , improve the query request and improve the accuracy rate of the query result . next , a personalized ranking algorithm is researched , a result feedback mechanism is added , a personalized search algorithm aiming at the mobile user is given in combination with the model , the ranking of the search results is optimized , the personalized domain retrieval service output can be given for the search input of different users , the webpage of interest of the user is placed in front , the search accuracy can be effectively improved , and the user interest model is updated according to the relevant feedback mechanism . Finally , using the characteristics of autonomy , learning and communication coordination of intelligent agent to design mobile personalized search engine prototype , a large and complicated system is divided into a plurality of independent autonomous agent modules , and a core module is developed to simulate the simulation experiment . The experimental results show that the accuracy of the retrieval result is better than that of the traditional retrieval result after the context information extension algorithm is added , and the satisfaction of the user to the improved system is improved by 10 % -25 % compared with the prior retrieval result .

【学位授予单位】：湖南大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.3

【参考文献】