XTQA: Span-Level Explanations of the Textbook Question Answering

TL;DR

This paper introduces XTQA, a novel model that provides span-level explanations for textbook question answering, enhancing interpretability and accuracy by identifying relevant evidence spans in multi-modal contexts.

Contribution

We propose a coarse-to-fine algorithm for span-level explanations in TQA, improving interpretability and state-of-the-art performance.

Findings

Significant performance improvement over baselines

Effective span-level evidence extraction

Enhanced explainability for TQA tasks

Abstract

Textbook Question Answering (TQA) is a task that one should answer a diagram/non-diagram question given a large multi-modal context consisting of abundant essays and diagrams. We argue that the explainability of this task should place students as a key aspect to be considered. To address this issue, we devise a novel architecture towards span-level eXplanations of the TQA (XTQA) based on our proposed coarse-to-fine grained algorithm, which can provide not only the answers but also the span-level evidences to choose them for students. This algorithm first coarsely chooses top $M$ paragraphs relevant to questions using the TF-IDF method, and then chooses top $K$ evidence spans finely from all candidate spans within these paragraphs by computing the information gain of each span to questions. Experimental results shows that XTQA significantly improves the state-of-the-art performance compared with baselines. The source code is available at https://github.com/keep-smile-001/opentqa