Large Language Models Meet Knowledge Graphs to Answer Factoid Questions

TL;DR

This paper introduces a method that combines knowledge graphs with large language models to improve factoid question answering by extracting relevant subgraphs and re-ranking answer candidates, achieving significant accuracy gains.

Contribution

It presents a novel approach for integrating knowledge graph subgraphs with pre-trained language models to enhance factoid question answering performance.

Findings

Re-ranking with extracted subgraph information improves Hits@1 scores by 4-6%.

The method effectively utilizes subgraph linearization for better answer candidate evaluation.

Knowledge graph integration enhances the interpretability and accuracy of language model-based QA.

Abstract

Recently, it has been shown that the incorporation of structured knowledge into Large Language Models significantly improves the results for a variety of NLP tasks. In this paper, we propose a method for exploring pre-trained Text-to-Text Language Models enriched with additional information from Knowledge Graphs for answering factoid questions. More specifically, we propose an algorithm for subgraphs extraction from a Knowledge Graph based on question entities and answer candidates. Then, we procure easily interpreted information with Transformer-based models through the linearization of the extracted subgraphs. Final re-ranking of the answer candidates with the extracted information boosts Hits@1 scores of the pre-trained text-to-text language models by 4-6%.