An integrated method of hotel site selection based on probabilistic linguistic multi-attribute group decision making

Abstract

The site selection acts a pivotal role in determining the success of a construction project. Since the site selection needs to gather the wisdom of a group of decision makers (DMs) and involves many factors, it can be regarded as a multi-attribute group decision making (MAGDM) problem in artificial intelligence. The assessments of alternatives on attributes are expressed by probabilistic linguistic term sets (PLTSs). A new two-stage normalization method is proposed for PLTSs considering the psychological states of decision makers. A new score function for PLTS is defined. The best and worst method is extended for fuzzy preference relation. The individual objective attribute weights are determined via information entropy. The individual subjective attribute weights are derived through the extended best and worst method. The individual comprehensive attribute weights are derived by minimum relative entropy principle. The weights of DMs are acquired through an optimization model. It minimizes the deviation between the opinions of all DMs and the deviation between the individual and collective comprehensive attribute weight vectors, simultaneously, which effectively overcomes the drawback of only minimizing single deviation. A new method is presented for MAGDM with PLTSs. A hotel site selection example is demonstrated and comparative analyses are executed to verify the validity and advantages of the proposed method. The test statistic $Z$ values of Spearman's rank-correlation test are all smaller than 1.645, which shows that the ranking order obtained by the proposed method is statistically sharply distinct from that produced by other methods and thus further validates the proposed method.