Problem
Research questions and friction points this paper is trying to address.
Employment-Discrimination Complaints
Event Knowledge Graphs
Complex Event Sequences
Text Representation
ARGUS: Role-Aware Event Knowledge Graphs for U.S. Employment-Discrimination Complaints
🤖 AI Summary
This study addresses the challenge that complex event sequences in U.S. employment discrimination complaints are inadequately captured by traditional lexical or embedding-based representations. To this end, it proposes an Event Knowledge Graph (EKG) construction paradigm that integrates structured generation with multi-granularity merging. Specifically, the method designs a source-grounding pipeline based on a 5W1H heuristic schema, synergizing domain-specific legal models with large language models for structured information extraction to generate document-level EKGs that consolidate participants, temporal dynamics, and causal relations, thereby enhancing evidence organization and reasoning capabilities. Experimental results demonstrate that graph-structured classifiers significantly outperform baseline models, and EKG retrieval effectively improves bounded question-answering performance, validating its practical value in legal evidence reasoning.
📝 Abstract
U.S. employment-discrimination complaints describe complex event sequences that are not explicitly captured by lexical or embedding-based representations alone. We present ARGUS, a source-grounded pipeline that combines a 5W1H-inspired schema, legal-domain models, and LLM-based structured generation to construct document-level Event Knowledge Graphs (EKGs) from CourtListener complaints. ARGUS extracts fact-bearing statements, builds chunk-level event graphs with participant, temporal, and causal structure, and merges them into document-level representations. We evaluate graph quality through human and multi-model assessment and test downstream utility on claim classification and legal QA. The graph-structured classifier outperforms raw and linearized baselines on the held-out set, and EKG-only retrieval improves document-scoped QA, while open-retrieval gains remain limited by low first-stage candidate recall. These results suggest that EKGs are most useful for organizing and reasoning over evidence once relevant material has been retrieved.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Event Knowledge Graphs
Large Language Models
Legal NLP
Structured Generation
5W1H Schema
🔎 Similar Papers
No similar papers found.
