基于情境和主体特征融入性的多维度个性化推荐-信息通信学术期刊网.PDFVIP

第33 卷第Z1 期 通 信 学 报 Vol.33 No. Z1 2012 年9 月 Journal on Communications September 2012 doi:10.3969/j.issn.1000-436x.2012.z1.003 基于情境和主体特征融入性的多维度个性化推荐模型研究 琚春华，鲍福光 (浙江工商大学计算机与信息工程学院，浙江 杭州 310018 ） 摘 要：个性化推荐准确率的高低是互联网应用成功与否的关键因素，针对传统推荐模型的不足，提出一种基于 情境和主体特征融入性的多维度个性化推荐模型，该模型能够充分利用地域文化背景、领域主题情景、主体特征 等信息，避免了传统算法把用户整体作为单个向量的弊端，克服了数据稀疏性等问题。实验结果表明，该模型的 推荐质量比传统的协同推荐模型高，更有针对性地向用户推荐他们感兴趣的项目。 关键词：推荐模型；个性化；多维度；情境；特征选取 TP393 A 1000-436X(2012)Z1-0017-11 中图分类号： 文献标识码： 文章编号： Research on a multidimensional personalized recommendation model based on a situation and characteristics of the users JU Chun-hua, BAO Fu-guang (School of Computer Science and Information Engineering, Zhejiang Gongshang University, Hangzhou 310018, China) Abstract: The accuracy of personalized recommendation was the key factor of Internet application to success. Because of the deficiency of the traditional recommend model, a multidimensional personalized recommendation model based on a special situation and main characteristics of the users was proposed. This model could make full use of regional culture background, field scene, characteristics of the users and so on, avoided the disadvantages of traditional algorithm, put the users overall characteristics as a single vector, and overcomed the problem of sparse data. The experimental results show that the quality of this recommendation model is better than traditional collaborative recommend model with the more specific items match user interests. Key words: recommend model; personalized; multidimens