Explaining Answers with Entailment Trees

TL;DR

This paper introduces a novel approach to explain open-domain QA answers using entailment trees, supported by a new dataset, ENTAILMENTBANK, enabling models to generate multi-step reasoning explanations.

Contribution

It presents the first dataset of multistep entailment trees and demonstrates that language models can partially generate these explanations, advancing interpretability in QA systems.

Findings

35% of generated trees are perfect when relevant sentences are provided

Models show some generalization to other domains

New dataset enables systematic explanation generation

Abstract

Our goal, in the context of open-domain textual question-answering (QA), is to explain answers by showing the line of reasoning from what is known to the answer, rather than simply showing a fragment of textual evidence (a "rationale'"). If this could be done, new opportunities for understanding and debugging the system's reasoning become possible. Our approach is to generate explanations in the form of entailment trees, namely a tree of multipremise entailment steps from facts that are known, through intermediate conclusions, to the hypothesis of interest (namely the question + answer). To train a model with this skill, we created ENTAILMENTBANK, the first dataset to contain multistep entailment trees. Given a hypothesis (question + answer), we define three increasingly difficult explanation tasks: generate a valid entailment tree given (a) all relevant sentences (b) all relevant and some irrelevant sentences, or (c) a corpus. We show that a strong language model can partially solve these tasks, in particular when the relevant sentences are included in the input (e.g., 35% of trees for (a) are perfect), and with indications of generalization to other domains. This work is significant as it provides a new type of dataset (multistep entailments) and baselines, offering a new avenue for the community to generate richer, more systematic explanations.