Knowledge Fusion and Semantic Knowledge Ranking for Open Domain Question Answering

TL;DR

This paper introduces a semantic knowledge ranking and fusion approach that enhances open domain question answering by improving knowledge retrieval and integration in BERT-based models, leading to better performance on QASC and OpenBookQA datasets.

Contribution

It proposes a novel semantic knowledge ranking model and a knowledge fusion model that leverages external knowledge with BERT, improving answer accuracy in open domain QA.

Findings

Knowledge fusion model outperforms previous methods.

Semantic re-ranking improves retrieval quality.

Enhanced BERT models achieve higher accuracy on QA datasets.

Abstract

Open Domain Question Answering requires systems to retrieve external knowledge and perform multi-hop reasoning by composing knowledge spread over multiple sentences. In the recently introduced open domain question answering challenge datasets, QASC and OpenBookQA, we need to perform retrieval of facts and compose facts to correctly answer questions. In our work, we learn a semantic knowledge ranking model to re-rank knowledge retrieved through Lucene based information retrieval systems. We further propose a "knowledge fusion model" which leverages knowledge in BERT-based language models with externally retrieved knowledge and improves the knowledge understanding of the BERT-based language models. On both OpenBookQA and QASC datasets, the knowledge fusion model with semantically re-ranked knowledge outperforms previous attempts.