We present \textbf{StepGap}, a hybrid NLI-LLM decision tree that detects step-level evidence gaps in multi-hop QA and emits one of three typed labels: \textsc{Contradicted Claim} (CC), \textsc{Irrelevant Evidence} (IE), or \textsc{Missing Bridge} (MB), each tied to a concrete repair action. On 82 multi-hop questions (181 annotated steps, κ=0.704), StepGap reaches sF1=72.0, within the bootstrap confidence interval of an LLM-only baseline (70.1) but with a more decomposable structure: every StepGap stage \emph{hurts} F1 when removed, while three of four LLM-only removals \emph{improve} F1 -- a sign of \emph{competing-error cancellation}, where internal stages mask each other's errors. We further expose a \emph{Q-F1 trap}: question-level F1 is mechanically inflated by checkers that flag every step, making step-level F1 the necessary diagnostic. Used as a typed GRPO process reward, StepGap improves Qwen2.5-7B-Instruct Exact Match from 32.1±0.3 to 35.4±0.9 across three seeds, with the single-run comparison showing a +5.6 Avg EM gain over the matched Search-R1 GRPO reproduction.
@article{arxiv.2605.24733,
title = {StepGap: A Hybrid NLI-LLM Checker for Step-Level Evidence-Gap Detectionin Multi-Hop Question Answering},
author = {Yuelyu Ji and Zhuochun Li and Hui Ji and Daqing He},
journal= {arXiv preprint arXiv:2605.24733},
year = {2026}
}