Retrieval Augmentation for Commonsense Reasoning: A Unified Approach

TL;DR

This paper introduces RACo, a unified retrieval-augmented framework for commonsense reasoning that leverages a large-scale commonsense corpus and novel retrieval strategies, achieving state-of-the-art results.

Contribution

The paper presents a new unified framework, RACo, with a large commonsense corpus and training strategies, specifically designed to enhance commonsense reasoning tasks.

Findings

RACo significantly outperforms existing knowledge-enhanced methods.

Achieved new state-of-the-art results on CommonGen and CREAK benchmarks.

Demonstrated the effectiveness of retrieval-augmented approach in commonsense reasoning.

Abstract

A common thread of retrieval-augmented methods in the existing literature focuses on retrieving encyclopedic knowledge, such as Wikipedia, which facilitates well-defined entity and relation spaces that can be modeled. However, applying such methods to commonsense reasoning tasks faces two unique challenges, i.e., the lack of a general large-scale corpus for retrieval and a corresponding effective commonsense retriever. In this paper, we systematically investigate how to leverage commonsense knowledge retrieval to improve commonsense reasoning tasks. We proposed a unified framework of retrieval-augmented commonsense reasoning (called RACo), including a newly constructed commonsense corpus with over 20 million documents and novel strategies for training a commonsense retriever. We conducted experiments on four different commonsense reasoning tasks. Extensive evaluation results showed that our proposed RACo can significantly outperform other knowledge-enhanced method counterparts, achieving new SoTA performance on the CommonGen and CREAK leaderboards.