Abstract: Aiming at the data sparsity problem of traditional recommendation method,common methods are limited to the amount of data.Therefore,the grey correlation prediction method was used to calculate the correlation coefficient between the program score data to predict the vacancy score data.In terms of the cold start problem for new users,the similarity of users' interest features and the similarity of users' ratings based on trust cloud were considered.Then,the comprehensive similarity between users was calculated to recommend the appropriate solution to the new user.Finally,the method validation was carried out by taking the car scheme recommendation as an example.Effectiveness of the method was proved by comparing with the recommendation algorithms such as collaborative filtering and cloud model.

Key words: scheme recommendation, sparsity, cold start, grey correlation prediction, trust cloud

摘要： 针对传统推荐方法中的数据稀疏性问题,常用的方法通常受到数据量的制约,因此采用灰色关联预测法计算方案评分数据间的相关系数,以预测空缺的评分数据;针对面向新用户的冷启动问题,考虑用户兴趣特征相似度和基于信任云的用户对方案评分的相似性,计算用户间的综合相似度,将合适的方案推荐给新用户。最后,以汽车方案推荐为例进行方法验证,并通过与协同过滤,云模型等推荐算法进行对比,证明了该方法的有效性。

关键词: 方案推荐, 稀疏性, 冷启动, 灰色关联预测, 信任云