| With the development of social network,large-scale group decision making(LSGDM)is attracting more and more attention,and the social commerce recommendation based on LSGDM is also the focus of recommendation system development.Under the background of big data,the decision-making environment and behavior are extremely complex.It is of great theoretical value and practical significance to analyze the problem of LSGDM in complex situations and effectively combine the theory of LSGDM with the social commerce recommendation system.This article will mainly focus on LSGDM methods,using the interval type-2 fuzzy sets(IT2 FSs)theory,which has the capability to deal with uncertain information flexibly,combining with the social network related methods and techniques,to study social commerce recommendation model and provide new ideas and solutions for the research of uncertain recommendation in complex situations.The main research results of this thesis are listed as follows:(1)In view of the uncertainty in the decision-making problem,the LSGDM method is studied from the aspects of preference expression and cluster analysis based on the IT2 FSs theory.Based on the data collected by questionnaire,the codebook of interval type-2 fuzzy linguistic variables is created for e-commerce comment situation.According to the large-scale characteristics of decision-making participants,an interval type-2 fuzzy equivalence relation(IT2 FERC)clustering algorithm is proposed,which presents the dynamic clustering effect according to different clustering levels.In addition,according to the large-scale characteristics of decision-making attributes,an interval type-2 fuzzy principal component analysis(IT2PCA)method is proposed to reduce the dimension of large-scale attributes.This method integrates similar attributes into new principal component attributes,which can effectively avoid the loss of attribute information in the clustering process.On this basis,the double LSGDM method which can deal with both large-scale decision-makers and attributes is proposed and applied to the analysis of online shopping behavior of e-commerce users.This research is the first time to carry out large-scale group decision-making research under the fuzzy environment of IT2 FSs,which provides a theoretical reference for the practical application of online commerce with language review.(2)Considering the social relationship among decision-makers,large-scale group decision-making methods based on IT2 FSs environment are studied.Based on the preference information expressed with interval type-2 fuzzy linguistic variables,the social relations among decision groups are identified.A LSGDM method based on IT2 FSs is proposed for the first time.The decision-making individuals are divided into corresponding sub-groups by community detection algorithm,and the weight of decision-making individuals and communities are calculated based on social network centralities.In addition,in view of the uncertainty of the expression of social relations,the strength of social relations is represented by interval type-2 fuzzy linguistic variables.A LSGDM method considering the combination of external social relations and internal preference relations under IT2 FSs environment is proposed.This research not only enriches the information of decision-makers through social relations to improve the quality of decision-making,but also uses social network analysis tools to greatly reduce the complexity of LSGDM.(3)According to the particularity and transitivity of trust relationship,the group decision theory and method based on trust relationship are studied.In view of the uncertainty of the expression of trust relationship,trust level is repreferented the interval type-2 fuzzy linguistic variables,the interval type-2 fuzzy trust propagation operator and aggregation operator are proposed according to trust propagative paths.Then,the implicit trust relationship is constructed for decision-makers through preference similarity.Based on this,the traditional minimum cost consensus model is improved to study the effect of trust relationship on group consensus reaching process(CRP).The subjective unit adjustment costs of decision-makers are modified based on trust relationship.In addition,according to the relationship contradictions in the field of decision-making,a multilayer network is constructed based on trust network and consensus evolution network,and the positive and negative effects of trust on consensus are studied.The above research content shows the promotion of trust to consensus in the way of model,and studies the possible negative effect of trust in group consensus from the perspective of contradiction,which promotes the in-depth and perfect research of group decision-making based on trust relationship.(4)According to the complexity of LSGDM problem,the CRP of LSGDM is studied from the perspective of cluster analysis.To balance the contradiction between clustering analysis and CRP,a dynamic clustering algorithm based on Lovain community detection method is proposed based on consensus evolution network,and the effective clustering results are obtained from the CRP by increasing consensus threshold.In addition,to reduce the complexity of interaction,the decision-maker's unit adjustment cost is considered as an auxiliary element of the preference relation in the cluster analysis process,so that sub-group members may have similar preferences and adjustment costs,which can reduce the difficulty of negotiation and save decision-making time.The above reseaches studies the consensus problem of LSGDM from the perspective of contradiction and clustering elements,which provides a new idea for the large-scale group consensus research.(5)To meet the needs of social commerce recommendation,a LSGDM method based on trust relationship is proposed,and its application in social commerce recommendation is studied.According to the transmission characteristics of trust relationship,individual decision-makers are divided into leaders and followers,and the initial division of trust community is obtained by taking leaders as the center.The trust propagation path from followers to corresponding leaders is determined,and the trust propagation operator and aggregation operator are used to evaluate the trust levels of followers to leaders.Based on this,followers belonging to more than one community are assigned to the unique community to get the final independent trust community.On this basis,combined with the recommendation model based on association rules,a social commerce recommendation model is proposed based on the trust-based LSGDM method,and it is applied to the movie recommendation scene through Movielens data analysis.The above research content combines the theory of LSGDM with the application of social commerce recommendation,realizes the transformation from theory to application,and provides a theoretical reference for promoting the development of online business. |
PDF Full Text Request