Semantic Reasoning with Differentiable Graph Transformations
This addresses the challenge of integrating logical reasoning with differentiable models for AI systems, though it appears incremental as it builds on existing graph and embedding methods.
The paper tackles the problem of semantic reasoning by introducing a differentiable reasoner that uses graph transformations as rules, which can be manually written or inferred from training data, with rules expressible in a subset of Description Logic.
This paper introduces a differentiable semantic reasoner, where rules are presented as a relevant set of graph transformations. These rules can be written manually or inferred by a set of facts and goals presented as a training set. While the internal representation uses embeddings in a latent space, each rule can be expressed as a set of predicates conforming to a subset of Description Logic.