A Linguistic Possibility-Probability Aggregation Model for decision analysis with imperfect knowledge

Knowledge which is imperfect creates obstructions in the operational parameter setting (OPS) of the functions of decision making; however it is relatively inexpensive to obtain the data to fill the knowledge gap by taking subjective measures. The collection of human experience data through individual subjective assessments tends to be aggregated as being of more objective value. This paper proposes a Linguistic Possibility-Probability Aggregation Model (LPPAM) to address this problem of OPS. LPPAM includes a Compound Linguistic Ordinal Scales model as a rating interface to improve the quality of data collection. The CLOS model addresses the problems with consistency, such as in the linguistic choices, accuracy of linguistic representation of numbers and bias of rating dilemmas. LPPAM also contains a multi-expert and multi-attribute aggregation model, which is to derive meaningful arguments fitting the setting of operation parameters. Five algorithms are used for LPPAM. The significance of LPPAM is that it can be applied in group multi-criteria decision problems and large scale survey systems. To validate the proposed framework, with a comparison with the Analytical Network Process (ANP), a R&D project selection problem is given.