论文部分内容阅读
商务企业应用数据挖掘技术向潜在客户推荐产品。大多数推荐系统聚焦研究兴趣于特定的领域,如电影或书籍。使用用户相似度或产品相似度的推荐算法通常可以达到较好效果。然而,当面临其他领域问题时,推荐常变得非常困难,因为数据过于稀疏,难以仅基于购买历史发现用户或产品间的相似性。为解决此问题,提出使用社会网络数据,通过对历史的观察提高产品推荐有效性。利用人工协同过滤和基于社会网络的推荐算法的最新进展进行领域推荐工作。研究显示社会网络的应用对于产品推荐具有很强的指导作用,但是,高的推荐精度需以牺牲召回率为代价。数据的稀疏性意味着社会网络并不总是可用,在这种情况下提出一种解决方案,很好地利用了社会网络的有效信息。
Business enterprises use data mining technology to recommend products to potential customers. Most recommender systems focus research interests in specific areas such as movies or books. Recommended algorithms that use user similarity or product similarity generally achieve better results. However, recommendations often become very difficult when faced with issues in other areas because data are too sparse to find similarities between users or products based solely on the buying history. In order to solve this problem, we propose using social network data to improve the effectiveness of product recommendation through the observation of history. Domain recommendation is made using the latest advances in manual collaborative filtering and social network-based recommendation algorithms. Studies have shown that the use of social networks has a strong guiding role for product recommendation, however, the high recommendation accuracy comes at the expense of the recall rate. The sparsity of data means that social networks are not always available, and a solution is proposed in this context, making good use of the information available in social networks.