基于双信息源的协同过滤算法研究

Research on dual information source model-based collaborative filtering algorithm

为了解决数据稀疏性,针对具有专门知识背景和交互性强的项目推荐,文章提出了基于双信息源模式的协同过滤算法,该方法判断活动用户对目标项目的兴趣程度建立在相似用户推荐组(最近邻居集合)与专家推荐组基础上,把2个推荐组的建议结合起来,形成可靠的信息源;并分析各自影响活动用户对目标项目的权重,计算活动用户的最终兴趣度,实现系统推荐。

This paper proposes Dual Information Source Model-Based Collaborative Filtering Algorithm（DISCF） in view of the data sparsity and interactive project recommendation with special knowledge.This algorithm determines the users＇ interest degree according to two recommendation groups,namely similar users＇ group and expert users＇ group and integrates them as credible information sources.Then,it analyzes the influence of each group on target users＇ weight of target product and calculates the ultimate interest degree of users to fulfill the systematic recommendation.