Trust modeling based on probabilistic linguistic term sets and the MULTIMOORA method

摘要

Trust modeling has attracted wide attention in different domains and served as the basis for decision-making under different contexts. Building a robust and effective trust model that considers various trust-related characteristics remains an enormous challenge. This report proposes a multi-criteria group decision-making based (MCGDM) trust evaluation model. We generalize our work from three aspects, which are trust metric, trust evaluation, and decision-making. First, we utilize the extension of the hesitant fuzzy linguistic term sets (HFLTS), called probabilistic linguistic term sets (PLTS), as our trust scaling method, which is a very suitable tool to describe the decision maker's opinions when they are hesitant about their judgments and intend to depict their evaluation information using several linguistic terms with corresponding probability. Second, In the process of trust evaluation, trust is decomposed into multiple trustworthiness facets with different importance degrees defined by the trustor, and a structural evaluation framework is established to evaluate the trustworthiness of each alternative. The unique properties of trust are also considered comprehensively. Specifically, the properties involve the subjectivity and context-sensitivity of trust in particular application scenarios, the hesitancy and uncertainty of decision-makers in expressing their assessment opinions, similarity between the trustor and the recommenders, and the dynamic reliability of the provided opinions. Finally, in the decision-making process, we adopt the Multi-Objective Optimization by Ratio Analysis (MULTIMOORA) method, which is a robust decisionmaking method that simultaneously fuses three subordinate orders to derive the final ranking. The experimental results demonstrate the effectiveness and accuracy compared with the other method.