Can KBQA Models Predict Their Reasoning Paths? Isomorphism Prediction Task as a Proxy

ACL ARR 2025 February Submission6748 Authors

Abstract: Despite achieving correct answers, we find that existing Knowledge Base Question Answering (KBQA) models struggle to follow the expected reasoning structures. We introduce the task of isomorphism prediction to enhance reasoning fidelity beyond answer generation, with a focus on generalization. We propose a contrastive knowledge co-distillation framework that unifies textual and graphical KBQA paradigms, improving isomorphism prediction and overall model generalization. Furthermore, incorporating isomorphism prediction as an auxiliary task could also improve KBQA performance.

Paper Type: Short

Research Area: Question Answering

Research Area Keywords: knowledge base QA, reasoning, generalization

Contribution Types: Model analysis & interpretability

Languages Studied: English

Submission Number: 6748