基于本地插件的个性化搜索研究

PERSONALIZED SEARCH RESEARCH BASED ON LOCAL ADD-ON

个性化搜索是一种通过采集、预测用户的兴趣爱好,根据用户的兴趣模型优化搜索结果的新型搜索形式。但传统的个性化搜索采用的兴趣模型不够准确,所提出的动态调整策略不够完备,不能够准确反映用户的兴趣变化。针对这些问题,提出了以二维兴趣多边形作为兴趣模型,基于本地插件的个性化搜索方案。详细介绍了二维兴趣多边形的建模、动态调整策略产生的动机和算法、网页领域模型的建立及适用于兴趣多边形的相似度匹配算法,最后通过实验验证了该个性化搜索方案能够提高搜索质量。

Personalised search is a new kind of search pattern which gathers and predicts users＇ interests and then optimises the search results according to users＇ interest model.However,the interest model used by traditional personalised search is not precise enough and the dynamic modification strategies are not perfect enough as well and can not precisely reflect the variations of users＇ interests.To achieve a good solution on these problems,a personalised search mechanism is proposed,which takes two-dimension interest polygon as the interest model and is based on the local add-ons.The details of modelling the two-dimension interest polygon,the motivation and algorithm of dynamic modification strategies,the generation of web page＇s domain model and the similarity matching algorithm fitting the interest polygon are systematically described in the paper.Finally,an experiment is performed and it proves that the personalised search mechanism is capable of improving search quality.