Probabilistic knowledge bases are commonly used in areas such as large-scale information extraction, data integration, and knowledge capture, to name but a few. Inference in probabilistic knowledge bases is a computationally challenging problem. With this contribution, we present our vision of a distributed inference algorithm based on conflict graph construction and hypergraph sampling. Early empirical results show that the approach efficiently and accurately computes a-posteriori probabilities of a knowledge base derived from a well-known information extraction system.
展开▼