Programming with Personalized PageRank: A Locally Groundable First-Order Probabilistic Logic

TL;DR

This paper introduces a first-order probabilistic language that enables efficient local grounding using personalized PageRank, improving inference scalability and speed in probabilistic logic systems, especially for large databases.

Contribution

The paper presents a novel probabilistic language extension that allows approximate local grounding with small graphs, leveraging personalized PageRank for scalable inference.

Findings

Grounding time is independent of database size.

Supervised weight learning improves accuracy.

Parallelization speeds up learning by an order of magnitude.

Abstract

In many probabilistic first-order representation systems, inference is performed by "grounding"---i.e., mapping it to a propositional representation, and then performing propositional inference. With a large database of facts, groundings can be very large, making inference and learning computationally expensive. Here we present a first-order probabilistic language which is well-suited to approximate "local" grounding: every query $Q$ can be approximately grounded with a small graph. The language is an extension of stochastic logic programs where inference is performed by a variant of personalized PageRank. Experimentally, we show that the approach performs well without weight learning on an entity resolution task; that supervised weight-learning improves accuracy; and that grounding time is independent of DB size. We also show that order-of-magnitude speedups are possible by parallelizing learning.