Efficient Markov logic inference for natural language semantics
Islam Beltagy, Raymond J. Mooney · 2014
Using Markov logic to integrate logical and distribu-tional information in natural-language semantics results in complex inference problems involving long, compli-cated formulae. Current inference methods for Markov logic are ineffective on such problems. To address this problem, we propose a new inference algorithm based on SampleSearch that computes probabilities of com-plete formulae rather than ground atoms. We also intro-duce a modified closed-world assumption that signifi-cantly reduces the size of the ground network, thereby making inference feasible. Our approach is evaluated on the recognizing textual entailment task, and experi-ments demonstrate its dramatic impact on the efficiency of inference. 1