论文部分内容阅读
认为建立种子集引导用户评分是解决协同过滤推荐系统新用户冷启动问题的方法之一。尝试将关联度引入种子集的构建策略,提出基于多属性综合评价的种子集策略,并利用公开数据集MovieLens设计实验,模拟推荐系统的新用户环境,对比不同种子集策略的预测准确度和成功率。实验结果表明,在更符合实际推荐系统需求的少量种子集情况下,考虑种子之间的关联性可以改善推荐效果。
It is considered that establishing a seed set to guide user rating is one of the methods to solve the problem of cold start of new users in collaborative filtering recommendation system. At the same time, this paper attempts to introduce the relevance degree into the construction strategy of seed set, proposes a seed set strategy based on multi-attribute comprehensive evaluation, and designs the experiment using the public data set MovieLens to simulate the new user environment of recommendation system and compares the forecast accuracy and success of different seed set strategies rate. The experimental results show that considering the correlation between seeds can improve the recommendation effect under the condition of a small number of seed sets that are more in line with the actual recommendation system requirements.