GreaseLM: Graph REASoning Enhanced Language Models for Question Answering

TL;DR

GreaseLM is a novel model that effectively combines pretrained language models and graph neural networks to improve reasoning in question answering tasks across various domains.

Contribution

It introduces a multi-layer fusion approach that allows bidirectional information flow between language context and structured knowledge representations.

Findings

Outperforms existing models on CommonsenseQA, OpenbookQA, and MedQA-USMLE benchmarks.

More reliably answers questions requiring reasoning over context and knowledge.

Achieves superior performance even with smaller model sizes.

Abstract

Answering complex questions about textual narratives requires reasoning over both stated context and the world knowledge that underlies it. However, pretrained language models (LM), the foundation of most modern QA systems, do not robustly represent latent relationships between concepts, which is necessary for reasoning. While knowledge graphs (KG) are often used to augment LMs with structured representations of world knowledge, it remains an open question how to effectively fuse and reason over the KG representations and the language context, which provides situational constraints and nuances. In this work, we propose GreaseLM, a new model that fuses encoded representations from pretrained LMs and graph neural networks over multiple layers of modality interaction operations. Information from both modalities propagates to the other, allowing language context representations to be grounded by structured world knowledge, and allowing linguistic nuances (e.g., negation, hedging) in the context to inform the graph representations of knowledge. Our results on three benchmarks in the commonsense reasoning (i.e., CommonsenseQA, OpenbookQA) and medical question answering (i.e., MedQA-USMLE) domains demonstrate that GreaseLM can more reliably answer questions that require reasoning over both situational constraints and structured knowledge, even outperforming models 8x larger.