自适应推荐算法在电子超市个性化服务系统中的应用研究

Research on personalized service system in E-supermarket by using adaptive recommendation algorithm

为了满足电子超市中用户的个性化的服务需求,提出并实现了一种基于支持向量机的自适应推荐算法。首先,将用户模型按照层次化方式组织成领域信息和原子需求信息,考虑多用户同类信息需求。采用支持向量机对领域信息节点中的原子需求信息进行分类协同推荐,然后在针对每一领域信息节点中的原子信息需求进行基于内容的过滤。该算法克服了分别采用协同推荐和基于内容的推荐单一方法的缺点,大大提高了信息的查准率和查全率,尤其适合大规模用户群的信息推荐。该算法用于基于电子超市的个性化推荐服务系统(PRSSES)中,结果表明是有效的。

To meet the personalized needs of customers in E-supermarket, a new adaptive recommendation algorithm based on support vector machine was proposed. Firstly, user profile was organized hierarchically into fied information and atomic information needs, considering similar information needs in the group users. Support vector machine （SVM） was adopted for collaborative recommendation in classification mode, and then vector space model （VSM） was used for content-based recommendation according to atomic information needs. The algorithm had overcome the demerit of using collaborative or content-based recommendation solely, which improved the precision and recall in a large degree. It also fit for large-scale group recommendation. The algorithm was used in personalized recommendation service system based on E-supermarket （PRSSES）. The system could support E-commence better. The results manifested that the algorithm was effective.