文献类型：期刊文献

中文题名：一种融合情境因素的社会化信息推荐新方法

英文题名：A New Method of Social Information Recommendation by Combining Contextual Factors

作者：房小可[1]

第一作者：房小可

机构：[1]北京联合大学应用文理学院

第一机构：北京联合大学应用文理学院

年份：2015

卷号：0

期号：22

起止页码：105-111

中文期刊名：图书情报工作

外文期刊名：Library and Information Service

收录：CSTPCD;;北大核心:【北大核心2014】；CSSCI:【CSSCI2014_2016】；

语种：中文

中文关键词：情境;社会化媒体;信息推荐;内容过滤;协同过滤

外文关键词：context;social media;information recommendation;content filtering;collaborative filtering

摘要：[目的 /意义]针对目前融合情境因素的信息推荐方法大都存在推荐前的情境过滤(pre-filtering)和推荐后的情境过滤(post-filtering)所导致的价值信息流失问题,将情境因素融入到推荐过程中,实现基于用户-资源-情境的多维推荐。[方法 /过程]将情境因素融入推荐的过程中,动态挖掘在不同情境下用户兴趣的偏好,利用社会网络的相关指标赋予用户兴趣初始值,从空间距离的视角计算用户兴趣的权重,最后,借鉴内容过滤和协同推荐的思想实现用户的评分预测,进而按照用户的兴趣进行推荐。[结果 /结论]与以往二维推荐的实验比较表明,将情境因素融入到推荐过程中的方法在减少价值流失的基础上,能更为准确地揭示用户的兴趣,提高推荐质量,为存在社会关系的社会化媒体推荐服务提供借鉴。
[ Purpose/significance] Since the current information recommendation method integrating contextual fac- tors leads to the loss of information valuedue to excessive filtration. This papereompletes muhidimensional recommendation based on use-resource-context. [ Method/process ] For such problems, this paper integrated context factors into the recom- mending process, and completed users＇ interest miningunder different contexts. Firstly, an initial interest value was given using the index of soeialnetworks. Then the weight for users＇ interest was proposed from the perspective ofspatial distance. Finally, users＇ interest score prediction was implemented according toeollaborative recommendation, and the interest rec- ommendations for the users was completed. [ Result/conclusion] The comparison between the two - dimensional recom- mendation and the multidimensional recommendation shows that the recommendation integrating context factors can more accurately reveal the users＇ interest and improve the quality of recommendation, and provide a reference for social media recommendation service.