Agent技术在元搜索引擎中的应用研究

发布时间：2018-11-20 14:41

【摘要】：随着因特网的飞速发展,网络信息的含量急剧增长,网络信息获取的方式也多种多样。如何准确、高效地获取有价值信息成为研究人员和用户日益关注的话题。现有独立搜索引擎存在数据库覆盖范围小、查全率偏低的缺陷,元搜索引擎通过调用多个独立搜索引擎的方式扩大了检索的范围,在一定程度上提高了查全率。但由于独立搜索引擎返回大量重复冗余结果,导致元搜索引擎的结果显示代理负担加大、系统查准率降低、响应时间过长。为解决这一缺陷,本文引入Agent技术,充分利用Agent的代理性、智能性和自主性的特点,将Agent技术运用到元搜索引擎中,集合两者各自在web信息挖掘和信息检索方面的优势,提高元搜索引擎的查询性能和检索效率。本文主要工作内容总结如下： 1、介绍Agent和MAS的相关理论知识,以及元搜索引擎的概念、工作原理等知识。分析和总结目前国内外关于Agent技术在元搜索引擎中的应用现状,提出传统元搜索引擎存在的不足。 2、建立一种基于奖励机制的智能元搜索引擎系统模型。该系统模型为各成员搜索引擎分别创建相应的成员Agent,利用成员Agent收集成员搜索引擎的查询结果并做相应处理。 3、提出一种与系统模型相适应的成员搜索引擎调度策略——基于奖励机制的成员搜索引擎调度策略。该调度策略充分考虑影响元搜索引擎查询性能的各个要素,并按照一定的奖励机制对各成员搜索引擎的查询重要度排名,优先选择重要度最高的若干成员搜索引擎进行调度。 4、提出一种与系统模型相适应的查询结果合成策略——基于奖励机制的查询结果合成策略。该合成策略针对已调度成员搜索引擎的查询结果,根据查询结果与查询请求的综合匹配度等级值,对查询结果进行归并排序。 5、用KQML语言实现系统中各Agent间的协作通信,并对系统性能进行综合分析和对比验证,证明这种基于奖励机制的智能元搜索引擎,其查全率、查准率和响应时间都在一定程度上优于传统元搜索引擎。
[Abstract]:With the rapid development of Internet, the content of network information increases rapidly. How to obtain valuable information accurately and efficiently has become a topic that researchers and users pay more and more attention to. The existing independent search engine has the defects of small coverage of database and low recall rate. Meta search engine expands the range of retrieval by calling multiple independent search engines and improves the recall to a certain extent. However, the independent search engine returns a large number of duplicate redundant results, which leads to the increase of the burden of the display agent of the meta-search engine, the reduction of the system precision rate and the long response time. In order to solve this defect, this paper introduces Agent technology, makes full use of the characteristics of Agent's generation rationality, intelligence and autonomy, applies the Agent technology to the meta search engine, sets their respective advantages in web information mining and information retrieval. Improve the query performance and retrieval efficiency of meta-search engine. The main work of this paper is summarized as follows: 1. Introduce the related theoretical knowledge of Agent and MAS, as well as the concept of meta search engine, working principle and so on. This paper analyzes and summarizes the application status of Agent technology in meta search engine at home and abroad, and points out the shortcomings of traditional meta search engine. 2. Establish an intelligent meta-search engine system model based on reward mechanism. The system model creates the corresponding member Agent, for each member search engine separately and collects the query results of the member search engine by using the member Agent and does the corresponding processing. 3. A member search engine scheduling strategy based on reward mechanism is proposed, which adapts to the system model. The scheduling strategy fully considers the factors that affect the query performance of the meta search engine, and ranks the member search engines according to a certain reward mechanism, and gives priority to the most important member search engines. 4. A query result composition strategy based on reward mechanism is proposed, which is suitable for the system model. The synthesis strategy is aimed at the query result of the scheduled member search engine. According to the comprehensive matching degree between the query result and the query request, the query results are merged and sorted. 5. The cooperative communication among the Agent in the system is realized by using KQML language, and the system performance is analyzed and compared. It is proved that this intelligent meta search engine based on the reward mechanism has its recall rate. Precision rate and response time are better than traditional meta-search engine to some extent.
【学位授予单位】：河北工程大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.3;TP18

【参考文献】