基于混合算法的电子商务推荐系统设计研究

针对目前电子商务推荐系统不能很好地为未注册用户服务的问题，根据未注册用户和注册用户各自特点采用两套不同的数据收集方案，以提高网站的友好性与数据的准确性。在推荐算法方面，由于决策树归纳算法与贝叶斯网络算法各有优劣，故将两种算法结合使用，并引入基于内容的算法思想对商品的各属性进行研究，以提高推荐的准确性。实验证明：上述方法能够很好地为未注册用户服务且基于混合算法的推荐优于基于单种算法的推荐。

Design of an E-commerce Recommender System Based on Hybrid Algorithm

In view of the current E-commerce Recommender System can not be good for unregistered users, the paper set two different sets of data collection program according to the characteristics of unregistered users and registered users, in order to enhance the friendly of website and the accuracy of the data. Because the decision tree algorithm and bayesian network algorithms both have advantages and disadvantages, the paper uses a combination of two algorithms, and introduces the content-based algorithm to research the attribute of goods to improve the accuracy of the recommendation. The experiments prove that these methods can provide good service for unregistered users and the recommendation based on the hybrid algorithm is superior to single algorithm.