基于单值中智集的协同过滤推荐算法

为缓解推荐系统中用户模糊评价带来的推荐准确性低的问题,构建基于单值中智集的协同过滤推荐模型.首先,构建用户—项目评分矩阵,划分用户评分等级,并将用户评分按照单值中智计算公式转换得到评分对应的真值隶属度、不确定值隶属度、非真值隶属度.然后,引入极端评价计算公式,将其作为极端评分惩罚系数,得到基于单值中智数分数的相似度计算公式.最后,结合杰卡德相关系数得到最终用户相似度.基于单值中智集的协同过滤推荐方法在公开数据集MovieLens上比较验证,实验结果发现融合单值中智集的方法在RMSE、MAE评价指标上均比其他方法有2％～3％的提升,能够有效提高推荐精度,更好地处理模糊问题.

Collaborative filtering recommendation algorithm based on single-valued neutrosophic set

User fuzzy evaluation in recommendation system usually cause low recommendation accuracy, in order to alleviate the problem, this paper designed a collaborative filtering recommendation model based on single-valued neutrosophic set. Firstly, this paper constructed a user-item scoring matrix, divided user rating, and converted the user rating according to the single-valued neutrosophic set calculation formula in order to obtain the true value membership degree, uncertain value membership degree and non-true value membership degree respectively. Next, this paper introduced the calculation formula of extreme evaluation as the penalty coefficient of extreme rating, based on which the similarity calculation formula is designed according to the single-valued neutrosophic set score. Finally, this model obtained the final-user similarity by combining the Jaccard correlation coefficient. The proposed collaborative filtering recommendation method was compared with the traditional similarity recommendation methods on the public data set MovieLens. The experimental results show that the proposed method is improved about 2-3 percent in index of RMSE and MAE, and the recommendation accuracy is efficiently promoted to deal with the fuzzy problem.