集成语义信息的电子商务推荐系统

E-commerce recommendation system of integrated semantic information

针对现有电子商务推荐系统中存在的数据稀疏性问题和系统冷开始问题,结合语义相似性以及产品分类学的方法,提出建立产品的语义关联模型,该模型的结构通过一个行业语义信息训练中心的训练得到,最终产品的语义信息被提取出来集成到现有的基于项目的协同过滤方法中。研究结果表明该方法在一定程度上克服了稀疏性问题和冷开始问题。

There exists many problems in the recommended approaches in being,such as data sparsity problem,new-item problem and scalability problem.In this study,combining the semantic similarity with the product taxology,a product semantic relevance model is set up,the structure of this model is acquired by the training of an industry semantic training center,finally semantic information extracted from products is integrated with current hem-Based Collaborative Filtering approach.The results of this study indicate that the approach can resolve data sparsity problem and new-item problem to some extent.