Open-ended Commonsense Reasoning with Unrestricted Answer Scope

TL;DR

This paper introduces a method leveraging pre-trained language models to perform open-ended commonsense reasoning by iteratively retrieving reasoning paths from an external knowledge base, effectively handling large search spaces and implicit multi-hop reasoning.

Contribution

The work proposes a novel approach that uses iterative retrieval of reasoning paths with pre-trained models, eliminating the need for task-specific supervision in open-ended commonsense reasoning.

Findings

Achieves better performance on benchmark datasets

Effectively handles implicit multi-hop reasoning

Demonstrates qualitative improvements over existing methods

Abstract

Open-ended Commonsense Reasoning is defined as solving a commonsense question without providing 1) a short list of answer candidates and 2) a pre-defined answer scope. Conventional ways of formulating the commonsense question into a question-answering form or utilizing external knowledge to learn retrieval-based methods are less applicable in the open-ended setting due to an inherent challenge. Without pre-defining an answer scope or a few candidates, open-ended commonsense reasoning entails predicting answers by searching over an extremely large searching space. Moreover, most questions require implicit multi-hop reasoning, which presents even more challenges to our problem. In this work, we leverage pre-trained language models to iteratively retrieve reasoning paths on the external knowledge base, which does not require task-specific supervision. The reasoning paths can help to identify the most precise answer to the commonsense question. We conduct experiments on two commonsense benchmark datasets. Compared to other approaches, our proposed method achieves better performance both quantitatively and qualitatively.