ProPPR: efficient first-order probabilistic logic programming for structure discovery, parameter learning, and scalable inference
William Yang Wang, Kathryn Mazaitis, William W. Cohen · 2014
A key challenge in statistical relational learning is to develop a semantically rich formalism that supports efficient proba-bilistic reasoning using large collections of extracted infor-mation. This paper presents a new, scalable probabilistic logic called ProPPR, which further extends stochastic logic pro-grams (SLP) to a framework that enables efficient learning and inference on graphs: using an abductive second-order probabilistic logic, we show that first-order theories can be automatically generated via parameter learning; that in pa-rameter learning, weight learning can be performed using par-allel stochastic gradient descent with a supervised personal-ized PageRank algorithm; and that most importantly, queries can be approximately grounded with a small graph, and in-ference is independent of the size of the database.