内涵粗糙三支概念及个性化推荐

Three-way concept with rough intent and its application in personalized recommendation

作为一种高效的数据分析工具,三支概念分析为推荐系统提供了新的思路。为解决评分系统中存在的数据稀疏问题,该文提出一种内涵粗糙三支概念以及基于它的个性化推荐方案。内涵粗糙三支概念由三支概念扩展得来,是由外延、正内涵以及负内涵组成的三元组,正负内涵由外延结合对应的阈值α和β分别生成,内涵粗糙三支概念具有边界模糊、信息丰富等特点。首先,以最大化概念体积为优化目标,设计启发式方法针对每个用户生成相应的内涵粗糙三支概念;其次,移除重复的内涵粗糙三支概念以构成概念集;最后,基于正负内涵特性和推荐置信度,对用户进行个性化推荐。在6个公开数据集上进行实验,结果表明,与k NN、MF、IBCF、CFGAN以及基于形式概念的推荐算法GreConD-k NN和GRHC相比,该文算法具有更优的推荐效果。

As an efficient data analysis tool,three-way concept analysis provides new idea for the field of recommendation systems.To solve the data sparse problem in rating systems,this paper proposes a three-way concept with rough intent and a personalized recommendation algorithm based on it.The three-way concept with rough intent is a triple consisting of extent,positive intent and negative intent.The positive and negative intents are generated by the extent and the corresponding thresholdsαandβrespectively.Compared with the traditional formal concept,the three-way concept with rough intent has the characteristics of fuzzy boundaries and rich information.Firstly,with the optimization goal of maximizing concept volume,a heuristic method is designed to generate three-way concept with rough intent for users;secondly,remove duplicate concepts to construct concept sets;finally,design recommendation scheme based on positive and negative intent characteristics and recommendation confidence and give recommendation results.Experiments are conducted on 6 public datasets,and the results show that compared with k NN,MF,IBCF,CFGAN,and the formal concept-based recommendation algorithms GreConD-k NN and GRHC,the proposed algorithm has better recommendation effect.