Group Interactive Decision-Making Based On Social Network And Its Application In Recommender Systems

Social network group decision-making(SN-GDM)is one of the most significant parts of decision-making research.Owing to the development of various types of social media,we can more easily obtain users' social relationships and behavioral data.Intuition-based and empirical decision-making methods are gradually being replaced by data-driven approaches.With the increase in the number of decision-makers and alternatives,many group decision-making methods are faced with the problems of low accessibility of data and high computational complexity,which cannot be applied in actual decision-making scenarios in the era of big data.To solve the above problems,the research contents of this paper are as follows:(1)This paper simulates the group interaction process through the following-based social relationships in social media.Considering the mechanism of link analysis in web page ranking,the Page Rank,Trust Rank,and People Rank algorithms are used to calculate the social influence of decision makers(DMs)in SN-GDM.Page Rank can calculate the DMs' actual influence with the same initial influence.Trust Rank algorithm can raise the influence of trusted DMs and weaken the influence of potentially deceptive DMs.The People Rank algorithm reflects the Matthew effect.Based on the original Page Rank algorithm,it shows that ‘the actual influence of DM with large initial influence will increase,while the actual influence of DM with small initial influence will decrease'.(2)Aiming at the topic-irrelevance problem of the link analysis algorithm,the persona method is introduced into SN-GDM.DMs are clustered based on their label variables.The influence of DMs is normalized within different persona groups to obtain DMs' weights.Then,we calculate the evaluation value of each alternative based on the scores given by the DMs.Finally,each persona group gets a ranking result that meets their own preferences.(3)The case study uses questionnaires to obtain data on the preferences of 24 students in a certain university for movie recommendation.DMs are divided into three groups with different persona by hierarchical clustering method.Also,the movie scores and rankings calculated by the three algorithms are given respectively.The results show that the global ranking obtained by considering all DMs is significantly different from the groupdecision-making results divided by personas.The influence of DMs calculated by three different algorithms will have an impact on the final decision/recommendation result.The method proposed in this paper takes both the actual influence of DMs and their preferences into consideration.It not only further enriches the SN-GDM research theoretically,but also offers a reference for the application of recommender systems.