WP1.1 Survey of algorithms for non-monotonic formalisms applied to geographic information

According to the report on non-monotonic reasoning (Year 1 deliverable), three approaches has been presented: default reasoning, auto-epistemic logic and circumscription. The automation of these non-monotonic reasoning formalisms has been studied and several methods and theorem provers have been proposed, most of them dedicated to default reasoning. For default reasoning, several approaches have been proposed, truth maintenance based system [8], [7], tableaux-based theorem prover [18], constraint satisfaction problem [2], sequent-calculus based theorem prover [3], genetic algorithms [9], prolog technology theorem prover [19], translation of default logic into extended logic programs [11], [14]. Automating of autoepistemic logic has been studied with a truth maintenance based system [7], and in [12], [13]. The automation of circumscription has been proposed with [1], [6], based on linear resolution in [16], while compiling circumscription into logic programs [4], [5]. In the last decade, since logic programs with stable model semantics can be equivalently described in terms of reducing logic programs to default logic, autoepistemic logic or circumpscription, the knowledge representation community focused on logic programming as non-monotonic reasoning formalism. A lot of sessions in international conferences and workshops have been dedicated to this topic. Several algorithms and efficient systems have been developped and implemented [10, 11, 17, 15]. This is the reason why we focused the survey on this most recent and promising representation formalism. This approach is presented in the attached document “logic programming as a non-monotonic reasoning formalism”. In WP1.3 we show how logic programming can be used in the context of geographic information and we give an illustration with the Flooding Problem.