@inproceedings{zhang-etal-2026-fixed,
title = "What Should Be Fixed? Rule-Grounded Corrective Actions for Reworkable Logical Anomalies in Industrial Inspection",
author = "Zhang, Haitao and
Chen, Bosheng and
Xu, Haote and
Tu, Xiaotong and
Huang, Yue and
Ding, Xinghao",
editor = "Atanasova, Pepa and
Haffari, Gholamreza and
Wang, Minghan",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-industry.258/",
pages = "4349--4356",
ISBN = "979-8-89176-460-6",
abstract = "Logical anomalies in industrial inspection are rule violations: an object count is wrong, an item is misplaced, a component type does not match, or a required relation is broken. Existing anomaly understanding systems usually stop at diagnosis, answering whether an image is anomalous, where the anomaly is, or why it violates a rule. Reworkable inspection workflows require the next question: \textit{What should be fixed?} We introduce an industry-oriented benchmark and training pipeline for rule-grounded corrective actions. The task asks a model to produce an operator-facing action that supports one of three downstream decisions: no action, rework, or reject/manual review. This setting exposes a \textit{Diagnosis-Action Gap}: identifying an abnormal object does not imply knowing the valid, minimal action that restores normality. We construct from MVTec LOCO and from VID-AD, adding correction-oriented annotations that connect normal rules, visual evidence, repairability, corrective actions, and verification targets. The datasets support multiple-choice correction selection, structured JSON action prediction, and open-ended correction generation. We further train \textbf{ReworkVLM}, a Qwen3-VL-based model optimized with , which combines semantic correctness, trace-answer alignment, and minimal-edit rewards. Experiments on both benchmarks show that correction-oriented evaluation reveals failures hidden by diagnosis alone, including wrong action direction, unnecessary repair, and non-minimal edits, while ReworkVLM improves structured corrective success and open-ended correction quality over general and anomaly-specific VLM baselines."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-fixed">
<titleInfo>
<title>What Should Be Fixed? Rule-Grounded Corrective Actions for Reworkable Logical Anomalies in Industrial Inspection</title>
</titleInfo>
<name type="personal">
<namePart type="given">Haitao</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bosheng</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haote</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaotong</namePart>
<namePart type="family">Tu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinghao</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Pepa</namePart>
<namePart type="family">Atanasova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gholamreza</namePart>
<namePart type="family">Haffari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minghan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-460-6</identifier>
</relatedItem>
<abstract>Logical anomalies in industrial inspection are rule violations: an object count is wrong, an item is misplaced, a component type does not match, or a required relation is broken. Existing anomaly understanding systems usually stop at diagnosis, answering whether an image is anomalous, where the anomaly is, or why it violates a rule. Reworkable inspection workflows require the next question: What should be fixed? We introduce an industry-oriented benchmark and training pipeline for rule-grounded corrective actions. The task asks a model to produce an operator-facing action that supports one of three downstream decisions: no action, rework, or reject/manual review. This setting exposes a Diagnosis-Action Gap: identifying an abnormal object does not imply knowing the valid, minimal action that restores normality. We construct from MVTec LOCO and from VID-AD, adding correction-oriented annotations that connect normal rules, visual evidence, repairability, corrective actions, and verification targets. The datasets support multiple-choice correction selection, structured JSON action prediction, and open-ended correction generation. We further train ReworkVLM, a Qwen3-VL-based model optimized with , which combines semantic correctness, trace-answer alignment, and minimal-edit rewards. Experiments on both benchmarks show that correction-oriented evaluation reveals failures hidden by diagnosis alone, including wrong action direction, unnecessary repair, and non-minimal edits, while ReworkVLM improves structured corrective success and open-ended correction quality over general and anomaly-specific VLM baselines.</abstract>
<identifier type="citekey">zhang-etal-2026-fixed</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-industry.258/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>4349</start>
<end>4356</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T What Should Be Fixed? Rule-Grounded Corrective Actions for Reworkable Logical Anomalies in Industrial Inspection
%A Zhang, Haitao
%A Chen, Bosheng
%A Xu, Haote
%A Tu, Xiaotong
%A Huang, Yue
%A Ding, Xinghao
%Y Atanasova, Pepa
%Y Haffari, Gholamreza
%Y Wang, Minghan
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-460-6
%F zhang-etal-2026-fixed
%X Logical anomalies in industrial inspection are rule violations: an object count is wrong, an item is misplaced, a component type does not match, or a required relation is broken. Existing anomaly understanding systems usually stop at diagnosis, answering whether an image is anomalous, where the anomaly is, or why it violates a rule. Reworkable inspection workflows require the next question: What should be fixed? We introduce an industry-oriented benchmark and training pipeline for rule-grounded corrective actions. The task asks a model to produce an operator-facing action that supports one of three downstream decisions: no action, rework, or reject/manual review. This setting exposes a Diagnosis-Action Gap: identifying an abnormal object does not imply knowing the valid, minimal action that restores normality. We construct from MVTec LOCO and from VID-AD, adding correction-oriented annotations that connect normal rules, visual evidence, repairability, corrective actions, and verification targets. The datasets support multiple-choice correction selection, structured JSON action prediction, and open-ended correction generation. We further train ReworkVLM, a Qwen3-VL-based model optimized with , which combines semantic correctness, trace-answer alignment, and minimal-edit rewards. Experiments on both benchmarks show that correction-oriented evaluation reveals failures hidden by diagnosis alone, including wrong action direction, unnecessary repair, and non-minimal edits, while ReworkVLM improves structured corrective success and open-ended correction quality over general and anomaly-specific VLM baselines.
%U https://aclanthology.org/2026.emnlp-industry.258/
%P 4349-4356
Markdown (Informal)
[What Should Be Fixed? Rule-Grounded Corrective Actions for Reworkable Logical Anomalies in Industrial Inspection](https://aclanthology.org/2026.emnlp-industry.258/) (Zhang et al., EMNLP 2026)
ACL