蒋翀 费洪晓 张啸

摘 要: 针对目前主流搜索引擎个性化程度低的问题,通过分析用户的浏览行为和浏览内容来获取用户的兴趣类别以及关键词,用一组带权重的关键词组成的向量集来表示用户兴趣模型,利用更新算法对模型进行更新与优化。将用户兴趣模型与开源搜索引擎Nutch相结合,加入中文分词组件IKAnalyzer,实现了个性化搜索引擎。进行了传统搜索和个性化搜索对比实验,结果证明,Nutch个性化搜索引擎结果更符合用户兴趣。
关键词: 用户兴趣模型; 个性化; 搜索引擎; Nutch
中图分类号:TP393 文献标志码:A 文章编号:1006-8228(2015)09-26-03
Research of personalized search engine based on user profile
Jiang Chong1, Fei Hongxiao2, Zhang Xiao2
(1. Modern Education Technology Center, HunanWoman's Vocational University, Changsha, Hunan 410004, China;
2. School of Software, Central South University of China)
Abstract: In order to improve the degree of personalization for popular search engine, the user's interest categories and keywords were got by analyzing user's browsing behavior and content. User profile was represented by a vector set which consisted of a set of weighted keywords and updated by correlated algorithm. By embedding in user profile and IKAnalyzer, Nutch became a personalized search engine. Comparative experiments were carried out with the traditional search and the personalized search. The results show that, the personalized search engine got more relevant result with user interest than traditional research engine and was proved to be effective.
Key words: user profile; personalized; search engine; Nutch
0 引言
飞速发展的互联网在带给人们海量信息的同时,也产生了难以让用户快速准确获取有效信息的问题[1]。目前,占市场主导地位的搜索引擎查询结果仅仅跟用户输入的关键词有关,并未考虑在相同关键字中所隐藏的用户个性化需求。这一类的搜索引擎以自动抓取信息和自动排序查找为主要特征[2]。目前,主流的搜索引擎均未实现面向客户需求和兴趣的个性化搜索。在这种情况下,个性化搜索引擎的研究和发展逐渐兴起。在这一代的搜索引擎中,公认的应该具备的特征是个性化和智能[3]。
为了根据用户需求和兴趣产生搜索结果,搜索引擎需要以用户兴趣模型的构建为基础。本文中采用隐式反馈的方式,通过分析用户的浏览行为和浏览内容,获取用户的兴趣类别和关键词,用一组带权重的关键词组成的向量集表示用户兴趣,利用更新算法对模型进行优化,使用户模型的构建能在指导的条件下进行,实现智能化的搜索。……