摘要
社会化标签正被广泛的应用在网页文本的描述和分类方面,能够直接反映用户兴趣和商品特征,因而可用于个性化推荐系统中。在进行标签推荐时,需要考虑到标签间的关联度,而现有的标签关联度算法都是基于标签之间的共现关系或者直接基于语义词典,这些算法未考虑到标签与资源的相关性和资源与资源的相关性。本文提出了基于语义的标签关联算法,首先通过潜层狄利克雷分配模型和向量空间模型求得资源间的相关度,然后通过概率模型求得标签与资源的相关度,最后求得标签间的关联度。实验结果表明:基于语义的标签关联算法能够有效的提升社会化标签推荐系统的性能,与语义词典在语义上基本一致且能够实现语义词典未登录词的关联。
Social tagging is an increasingly popular way to describe and classify documents on the web and can directly reflect the user interest and commodity characteristics, which can be used in personalized recommendation system. During the tag recommended, you need to take into account the tag relevance. The tag relevance with the existing algorithms are based on the degree of co-existing relationship between the tags or directly based on the semantic dictionary. The algorithms do not take into account the correlation between tags and resources and resources associated with the resource. In this paper, a tag relevance algorithm based on semantic, first by the Latent Dirichlet Allocation （LDA） and the Vector Space Model （VSM） obtained the correlation between resources and by the probability model obtained the correlation tag and resources, and finally obtained the tag relevance. Experimental results show that tag relevance algorithm based on semantic can effectively enhance the social performance of tag recommendation system, is semantically consistent with semantic dictionary, and make out-of-vocabulary （00V） achieve semantic association.
出处
《软件》
2012年第12期136-138,共3页
Software
