利用服务器日志优化搜索引擎

Optimizing of search engine using server log

随着因特网信息量的日益增长,网络用户对搜索引擎的功能、智能化程度和检索结果有了更高的要求,希望搜索引擎能够提供更为准确、可靠与符合其个性化需要的检索结果.对支持向量机(SVM)分类基本原理进行深入细致的研究,发现SVM算法可以用来学习文档检索函数且效果显著;探讨了如何利用改进的SVM算法根据用户行为日志中的用户兴趣学习更新检索函数.利用现有技术构建了一个搜索引擎,验证结果表明,该算法能够分析站点用户行为的共性,得到满足用户需求的检索函数.

With the increasing amount of Internet information, web users have high expectations for search engine function, intelligence and search results and hope that search engines provide more accurate, reliable and individual search results. It is found through investigating the basic classification principles of Support Vector Machines （SVM） that the SVM algorithm can be applied to learning document retrieval function, and the effect is significant. It was discussed how to learn and update retrieval function using the modified SVM algorithm and based on user interest in behavior log. A search engine was constructed based on the present technology. The verifying result shows that the algorithm can analyzes commonness of user behaviors and obtain retrieval function to meet needs of users.