| LPMLN（Logic Programming with Markov Logic Networks）extends the language of An-swer Set Programming（ASP）by assigning a weight degree to each rule so that its stable models do not have to satisfy all LPMLNrules,which is rooted in the manner of Markov Logic Net-works（MLN）to handle the uncertainties and inconsistencies in knowledge representation and reasoning.Due to its expressibility,LPMLNcan be employed in several real world applica-tions.However,an LPMLNprogram is much harder to solve than its unweighted counterpart（an ASP program）,and only some preliminary solvers have been implemented so far,which is preventing further studies in both theoretical and practical sides.By using native solving and parallelization techniques,efficiency of calculating stable models can be improved to remedy this situation.However,native solving requires theoretical support and parallelization methods are only frameworks which is not enough for algorithm implementation.The main work of this thesis are listed as follows.1.For native solving:Proposing a grounding method using translation.Extending com-pletion and loop formulas to LPMLN.Designing a conflict-driven LPMLNsolving method based on completion and loop formulas.2.For parallelization methods:Proposing an algorithm which uses propagation to assist partition augmented subsets.Proposing an algorithm to calculate splitting sets for splittable programs and an algorithm to calculate core atom set for arbitrary LPMLNprograms.Proposing an algorithm to calculate independent divisible programs.Proposing an approach to combine the parallelization methods based on analysis and comparison of the three parallelization methods.3.Implementation of LPMLNsolvers based on methods above.And experiments on the LPMLNsolvers. |
