Multi-stage probabilistic linguistic matching method with the screening mechanism and individual preference relationship fusion
提出一种多阶段双边匹配决策系统，利用概率语言集成云贝叶斯网络处理指标依赖关系，并改进ORESTE方法获取偏好排序，结合筛选机制构建动态匹配模型，以提升匹配效率和质量，案例验证于二手房交易。
Considering the diversification of transaction information and complexity of matching environment, the original information with the language tendency plays an important role. Thus, under this circumstance, this paper gives a matching decision-making system under the probabilistic linguistic environment so as to improve the matching efficiency and quality. First, this paper demonstrates the multi-stage two-sided matching problem and illustrates the system frame. Then, for the multiple indicators, a probabilistic linguistic integrated cloud Bayesian network is constructed to present the dependency relationship, and determine the corresponding probability. It is known that the accuracy of agents’ preference information acts on the stability of final matching results, which may involve the strong ranking, agents’ psychological preferences and personal interests, etc. Thus, the improved ORESTE (organísation, rangement et Synthèse de données relarionnelles, in French) method is introduced to derive strong ranking and determine preference, indifference, and incomparability relation (PIR). Furthermore, this paper constructs the dynamic two-sided matching model considering the screening effect. Finally, a case study in second-hand house transaction is used to demonstrate the matching process. Simulation and comparison analysis validate its feasibility and effectiveness.