BiCon-Gate: Consistency-Gated De-colloquialisation for Dialogue Fact-Checking

TL;DR

This paper introduces BiCon-Gate, a semantics-aware gating mechanism for dialogue fact-checking that improves evidence retrieval and verification by effectively handling colloquial language through staged de-colloquialisation.

Contribution

The paper proposes a novel gated approach that selectively applies de-colloquialisation, enhancing fact-checking accuracy in dialogue systems over existing methods.

Findings

Improved retrieval and verification performance on DialFact benchmark.

Strong gains on SUPPORTS category compared to baselines.

Outperforms decoder-based LLM rewrite approaches.

Abstract

Automated fact-checking in dialogue involves multi-turn conversations where colloquial language is frequent yet understudied. To address this gap, we propose a conservative rewrite candidate for each response claim via staged de-colloquialisation, combining lightweight surface normalisation with scoped in-claim coreference resolution. We then introduce BiCon-Gate, a semantics-aware consistency gate that selects the rewrite candidate only when it is semantically supported by the dialogue context, otherwise falling back to the original claim. This gated selection stabilises downstream fact-checking and yields gains in both evidence retrieval and fact verification. On the DialFact benchmark, our approach improves retrieval and verification, with particularly strong gains on SUPPORTS, and outperforms competitive baselines, including a decoder-based one-shot LLM rewrite that attempts to perform all de-colloquialisation steps in a single pass.