【作者】 盛宪锋；

【导师】 山岚；

【作者基本信息】 北京化工大学 ， 计算机应用， 2004， 硕士

【摘要】 随着网络信息资源的迅速增加和实时更新,如何及时准确地获取自己所需的有用信息是现代网络信息检索技术需要解决的主要问题。本课题为在石油化工专业背景下,对基于元搜索引擎的专业式、专题式、个性化网络智能信息检索系统的研究,并且给出了其框架实现。解决了人们获取信息的广度问题,同时采用了人工智能的智能体新技术,实现了基于用户兴趣的智能化检索,保证了信息的可靠性,即精度问题。为石油化工专业下的用户在Web网络中检索所需的有效信息提供了较好的方案。本系统中,专业、专题检索主要是通过建立、维护专业同义词和专题词库来实现的,用户提交的检索关键词传递到元搜索引擎接口中,调用多个搜索引擎,其返回结果由页面自动分析器分别对其进行分析整理,从而取得有效的结果信息。个性化检索的实现,首先是通过知识表示中的框架表示法对用户的知识信息进行存储,并在用户使用系统的过程中由用户的检索兴趣和对检索结果的反馈信息对用户知识做动态更新。检索结果的过滤算法应用了向量空间模型算法,将检索结果和查询条件进行矢量化,通过计算其夹角大小来衡量文档的相似度,作为结果显示的标准之一,此外还考虑了权威网站信息、元搜索引擎返回结果的权重信息等因素。更多还原

【Abstract】 Nowadays the information in Internet is increasing rapidly and being updated in real time. How to acquire the needed information quickly and accurately is a main topic of the Web information retrieval technology.With the petroleum and chemistry background,this paper focus its research on a new specialized、catalogic and personalized intelligent Web information retrieval system based on meta-search engine,and gives a systematic implementation. This system provides an effective method to acquire info in Web for the petroleum and chemistry user. It increases the scope of acquiring information, and increases the precision of the search by integrating the agent technology of AI to realize the intelligent retrieval. In this system, the specialized and catalogic retrieval is realized by building the maintenance of specialized synonymous words and subject phrase. The meta-search engine interface receives the user’s retrieval keyword and dispatches it to the search engines. Then the result of the search engines was dealed with by the automatic analyser. Consequently some useful information is acquired. To implement the personalized <WP=6>retrieval, this system firstly store the user’s initial info into the database via the Frame Format, and the user’s knowledge repository can be modified automaticly by user’s retrieval information and the feedback information of the retrieval result. The filtration algorithm for the retrieval result is the VSM (Vector Space Model), at first transform the words of the retrieval result document and querying condition words into digital number by VSM algorithm, and then compute the angle of the vectors to appraise the similarity of the querying condition words and retrieval result document, furthermore, considering the factor of the authority Web site info and the weight of the retrieval result from the meta-search engine.更多还原