基于加权派系的个性化信息推荐研究  

Personalized Information Recommendation Based On Weighted Cliques

作　　者：熊回香[1] 李跃艳 

机构地区：[1]华中师范大学信息管理学院,湖北武汉430079

出　　处：《情报科学》2018年第1期67-74,共8页Information Science

基　　金：国家社会科学基金项目(12BTQ038)

摘　　要：【目的/意义】面对网络时代数据的海量性和无序性,为用户推荐个性化资源有利于增强用户间合作、提高知识的共享速度,对新知识的发现具有深远意义。【方法/过程】基于具有相同兴趣用户的聚合优于单纯的信息聚合,构建基于社会化标注系统的个性化推荐模型。通过引入社会网络中用户使用标签的频次来选择与用户关联显著的标签,并通过加权派系发现和聚合"小众"凝聚组群和相似标签集,进而为用户推荐优质资源,使其真正契合用户的个性化需求偏好。【结果/结论】结果表明模型能够有效实现信息的个性化推荐,消除单独聚类带来的粗糙数据集,并通过抓取豆瓣上的数据进行实证分析。[Purpose/significance] In the face of the massive and disorder of the data in the Internet era,it is helpful to en- hance the cooperation between users and improve the sharing speed of knowledge, which is of great significance to the dis- covery of new knowledge. [Method/process] Based on the aggregation of the users with the same interest is better than the simple information aggregation, this paper constructed a personalized information model which is based on the folksonomy system.This paper introduced label frequency with the user on social networks to select a significant label associated with the user, and through the weighted cliques to find and aggregate ＂niche＂ cohesive group and similar label set, and then rec- ommended for the high-quality resources for the user, so that it really fit the user＇s personalized needs preferences. [Result/ conclusion] The results show that the model can effectively realize the information personalized recommendation, and re- duce the rough dataset bring by separate clustering, and the empirical analysis is conducted by data grabbing from Douban.

关 键 词：“小众”凝聚组群 加权派系 协同过滤 个性化信息推荐 

分 类 号：G254.9[文化科学—图书馆学]