To improve personalized search, we need to increase the efficiency of personalization models using effective user profiles and ranking models. The ranking models improve accuracy by combining personalized and non-personalized models. In the personalized models, user profiles are used to re-rank the results, while in non-personalized models documents are ranked in the absence of user profile. A personalization metric able to estimate the potential for personalization can enable the selective application of personalization and improve the overall effectiveness of the search system. In this paper, a personalization fuzzy topic model (FTM) is proposed for integrating the topical user profile into the personalized web search. The topical user profile is built using the fuzzy logic in handling the uncertainty of the occurrence of all topics in a document, and the fuzzy c-means algorithm is used to retrieve the relevant topics. To evaluate the proposed model, the ranking results using the proposed Personalized-FTM are compared against personalization using the Latent Dirichlet Allocation model. The result reveals that the Personalized-FTM improves the Mean Reciprocal Rank and the Normalized Discounted Cumulative Gain by 7% and 5%, respectively, for all topic numbers.

为了改善个性化搜索，我们需要使用有效的用户个人资料和排名模型来提高个性化模型的效率。排序模型通过结合个性化和非个性化模型来提高准确性。在个性化模型中，使用用户配置文件对结果重新排序，而在非个性化模型中，在没有用户配置文件的情况下对文档进行排名。能够估计个性化潜力的个性化度量可以启用个性化的选择性应用并提高搜索系统的整体有效性。在本文中，提出了一种个性化模糊主题模型（FTM），用于将主题用户配置文件集成到个性化Web搜索中。使用模糊逻辑构建主题用户配置文件，以处理文档中所有主题出现的不确定性，并且使用模糊c均值算法检索相关主题。为了评估建议的模型，将使用建议的Personalized-FTM的排名结果与使用潜在Dirichlet分配模型的个性化进行比较。结果表明，对于所有主题编号，个性化FTM分别将平均倒数排名和标准化折现累积收益提高了7％和5％。