Research On Probabilistic Epistemic Logic Program

The representation and reasoning of incomplete knowledge is an important challenge knowl-edge engineering researchers are facing.Currently,incomplete knowledge is commonly rep-resented as default knowledge,epistemic knowledge and imprecise knowledge.Answer Set Program(ASP)is an important tool for the representation and reasoning of default knowledge.Epistemic logic program is a combination of ASP and auto-epistemic logic and introduced epis-temic modalities into ASP,which provides an approach to representing and reasoning with epis-temic knowledge.LPMLN is a combination of ASP and Markov Logic Networks(MLN),and make logic programming language more powerful in probabilistic representation and calcula-tion.This thesis designs a new logic programming language,PELP,which is a new probabilistic epistemic logic programming language based on LPMLN,to provide an alternative approach to solving the intractable problems with defaults,epistemic and imprecise knowledge.The main contributions of this thesis include:1)Proposing the syntax and semantic of PELP(Probabilistic Epistemic Logic Program)and defining its syntax and semantic;2)Propos-ing a basic algorithm for solving PELP programs and an optimization approach on program simplification;3)Providing its implementation of this algorithm using Clingo,which is a well-known ASP solver;4)Investigating the relation between PELP and the existing epistemic logic programming languages such as ASP17,EFLP and GI-log;5)Exploring the applications of PELP by modeling and solving the well-known problems:Monty Hall problem and confor-mant probabilistic planning problems.