Learning personalized individual semantics through the data of distributed linguistic preference relations: A two-stage method to support linguistic consensus reaching. (October 2022)

• Record Type: Journal Article
• Authors: Gao, Yuan; Fan, Sha; Hu, Zhineng; Li, Cong-Cong; Dong, Yucheng
• Is Part Of: Computers & industrial engineering. Volume 172:Part A(2022)
• Publication Date: 2022-10
• Subjects: Distribution linguistic preference relations -- Personalized individual semantics -- Linguistic group decision making -- Consistency -- Consensus; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2022.108581
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: We propose a two-stage consensus approach in a distributed linguistic context. We integrate a PISs learning phase in the two-stage consensus approach. Our model not only obtains dynamical PISs but also manages consistency and consensus. Abstract: Distribution linguistic preference relation (DLPR) is an effective tool to model linguistic preferences with multiple linguistic terms, and has widespread adoption in the linguistic group decision making (GDM), in which we commonly accept the existence of personalized individual semantics (PISs). Currently, the existing research on PISs is driven by a consistency-based method and assumes the fixed PISs for decision makers. However, decision makers may update their PISs during the consensus reaching process (CRP), which implies the importance of learning PIS from the preference data. To analyze the changing PISs in the CRP in a linguistic GDM problem with DLPRs, we put forward a PIS-learning-based two-stage consensus approach by considering all data of DLPRs given by decision makers. The PIS-learning-based two-stage consensus approach uses the consistency-driven and consensus-driven methodology to update the personalized numerical scales (PNSs), and guarantee the acceptable consistency and consensus simultaneously. Finally, we apply the proposed method to the financial technology selection. Moreover, a comparative analysis verifies the effectiveness of the PIS-learning-based two-stage consensus approach for supporting GDM with DLPRs.