With the evolution of mobile internet,people are expecting better recommendations while consuming digital multimedia products.Music,as a multimedia product with consumption highly susceptible to user mood,is particularly difficult to recommend.In this paper,a hybrid music-recommendation mechanism considering user's mood as a major factor is proposed.Current dominant mood of the user is deduced by applying time variations to Human Mood Regulation Model,which comes from modern psychology.Higher dimensions help to generate a variety of sophisticated moods by utilizing an Artificial Mood Model and an almost accurate mood of the respective user is approachable with user-feedback.The mechanism injects in or filters out appropriate results of the recommendations in line with the existing methods.The results of experiments show superior performance of the proposed mechanism against existing methods.%随着互联网和移动互联网的发展,越来越多的用户在消费数字音像制品时,希望从繁多的内容中得到更好的推荐.不同于普通商品,音乐的消费过程与用户的情绪关联十分密切.本文设计了一种融入用户情绪因素的综合音乐推荐方法.该方法利用现代心理学中人类情绪的自然节律模型,结合时间变化来推测用户当前的主导情绪.通过人工情绪模型来增加情绪的维度,以产生粒度更精细的情绪,并通过用户输入的训练,来逼近用户当前的真实情绪.在现有的音乐推荐方法基础上,加入或过滤出情绪分类与用户情绪吻合的推荐结果.实验结果验证了情绪因素会对音乐推荐过程产生影响,构建的融入用户情绪因素的综合音乐推荐方法,比其他不考虑情绪因素的推荐策略好且稳定,具有较好的实际应用效果.
展开▼
