ConstructCIE: A Dataset for Extracting Causal Information from Construction Accident Narratives

2026-08-06

AI Summary

This study addresses the challenge of extracting implicit, long-range, and dispersed causal information from construction accident narratives. To this end, it proposes the first hierarchical annotation framework tailored for causal inference in construction accidents and introduces ConstructCIE, a manually annotated dataset derived from OSHA reports that encompasses accident types, causal factors, sub-factors, and their supporting evidence spans. Systematic evaluation of sequence labeling models and instruction-tuned large language models—specifically JHE and IHE—reveals that current approaches perform well in accident type prediction and macro-level causal semantic recovery but still struggle with precise evidence span extraction. Among the evaluated models, JHE achieves superior performance on exact and soft-match metrics, while IHE occasionally attains higher keyword-level F1 scores.

Abstract

Construction accident narratives contain rich causal information, but the evidence is often implicit, long-span, and distributed. We introduce ConstructCIE, a manually annotated dataset for Causal Information Extraction from OSHA construction accident reports. The dataset uses a hierarchical schema for accident types, causal factors, sub-causal factors, and supporting evidence spans. We evaluate supervised sequence taggers and instruction-tuned LLMs in an end-to-end hierarchical extraction setting. Results show that most evaluated models achieve strong accident-type prediction and recover broad causal meaning but remain limited in precise span-level extraction. JHE generally achieves stronger exact and soft matching, while IHE sometimes achieves higher keyword F1. Error distributions vary by extraction strategy, but evidence-selection and span-boundary errors remain common. These findings show that reliable Causal Information Extraction for construction accidents requires stronger domain grounding and more accurate evidence extraction.

Problem

• Causal Information Extraction
• Construction Accident Narratives
• Evidence Span Extraction
• Implicit Causality
• Hierarchical Information Extraction

Innovation

• Causal Information Extraction
• Construction Accident Narratives
• Hierarchical Annotation Schema
• Evidence Span Extraction
• Instruction-tuned LLMs