摘要
文章提出一种基于本体和标签的个性化推荐模型,可以有效解决标签的非等级结构、多样性、模糊性所导致的标签间语义缺乏的问题,从而提高基于社会化标签的个性化推荐效果。将预处理后的社会化标签映射到Word Net中,利用Word Net语义相似度算法计算成功映射的标签的语义,用统计学的方法计算不能成功映射的标签的语义,然后将标签自身频率和标签语义相结合计算用户标签权重,进而计算用户标签特征向量和资源标签特征向量的相似度,最后实现个性化推荐。实验表明,该方法优于传统的基于社会化标签的推荐。
This paper presents a personalized recommendation model based on ontology and tags,which can effectively solve the problem of semantic absence among tags due to the diversity,ambiguity and lack of hierarchy,so as to improve the effect of personalized recommendation based on social tag. The paper matches preprocessed tags to the Word Net and computes the semantics of the successful mapped tags by Word Net semantic similarity. As for the tags that cannot be matched successfully,the paper uses statistical methods to calculate the semantics. With the combination of the frequency of tags and the semantics of the tags,the paper first calculates the weight of user tags and then the similarity of user tag eigenvector and resource tag eigenvector in order to realize the personalized recommendation. Experimental results show that the method is superior to the traditional recommendation based on social tag.
出处
《情报理论与实践》
CSSCI
北大核心
2016年第12期114-119,共6页
Information Studies:Theory & Application
基金
国家自然科学基金项目"社会化媒体集成检索与语义方法研究"的成果之一
项目编号:71273194
关键词
本体
社会化标签
个性化推荐
ontology
social tag
personalized recommendation
作者简介
唐晓波，男，1962年生，博士，博士生导师。
钟林霞，女，1991年生，硕士生。
王中勤，女，1991年生，硕士生。
