【摘要】 根据用户的历史行为信息向用户推荐符合其偏好的商品列表是推荐系统的基本方法之一,用户的行为信息可能是显式的(如电子商务网站中的商品评分),也可能是隐式的(如点击商品详情信息或收藏商品).但在实际场景中,用户隐式行为数量往往多于显式行为数量.为了得到更为准确的推荐结果,本文引入用户经历来定义用户在系统中隐式反馈的累积数量,提出了一种利用用户经历作为平衡系数来平衡多种策略的自适应推荐模型(User Experience based adaptive Recommendation Model,UERM),然后通过引入阻尼系数对模型进行进一步优化,提出了融入阻尼系数的融合用户经历的自适应推荐模型(UERM+).最后在两个真实数据集上进行实验,证明本文模型能够有效地提高推荐精度.更多还原

【Abstract】 It is one of the basic methods of recommending a system to recommend a product list according to the user’s historical behavior information according to the user’ s historical behavior information. The user’ s behavior information may be explicit( such as product rating in an e-commerce w ebsite),or may be implicit.( such as clicking on a product listing or a favorite item). How ever,in actual scenarios,the number of implicit behaviors of users is often more than the number of explicit behaviors. In order to get more accurate recommendation results,this paper introduces user experience to define the cumulative amount of implicit feedback of users in the system,and proposes an adaptive recommendation model that uses user experience as a balance factor to balance multiple strategie.The Adaptive Recommendation M odel( UERM) is then further optimized by introducing the damping coefficient. A user-based adaptive recommendation algorithm( UERM +) incorporating the damping coefficient is proposed. Finally,experiments on tw o real data sets prove that the model can effectively improve the recommendation accuracy.更多还原