Abstract
Probabilistic Logic Programming is an active field of research, with many proposals for languages, semantics and reasoning algorithms. One such proposal, Logic Programming with Annotated Disjunctions (LPADs) represents probabilistic information in a sound and simple way. This paper presents the algorithm "Probabilistic Inference with Tabling and Answer subsumption" (PITA) for computing the probability of queries. Answer subsumption is a feature of tabling that allows the combination of different answers for the same subgoal in the case in which a partial order can be defined over them. We have applied it in our case since probabilistic explanations (stored as BDDs in PITA) possess a natural lattice structure. PITA has been implemented in XSB and compared with ProbLog, cplint and CVE. The results show that, in almost all cases, PITA is able to solve larger problems and is faster than competing algorithms.
Original language | English (US) |
---|---|
Journal | CEUR Workshop Proceedings |
Volume | 616 |
State | Published - Jan 1 2010 |
Event | 17th RCRA 2010 Workshop on Experimental Evaluation of Algorithms for Solving Problems with Combinatorial Explosion - Bologna, Italy Duration: Jun 10 2010 → Jun 11 2010 |
ASJC Scopus subject areas
- General Computer Science