Argus builds event graphs from legal complaints to improve claim classification
ARGUS: Role-Aware Event Knowledge Graphs for U.S. Employment-Discrimination Complaints
Computation and Language
Summary
Employment-discrimination complaints in the U.S. often describe complicated sequences of events that are hard to understand with basic text analysis tools. The authors created ARGUS, a system that extracts detailed event information—including who did what, when, and why—from these complaints and organizes it into structured graphs. These graphs help computers better understand the complaint details and improve tasks like classifying claims or answering legal questions. The research shows that this graph-based approach works better than simpler methods after relevant information is found.
What this means in practice
- •For legal technology teams: Enhance automated claim classification and legal question answering by using event graphs structured from court complaints.
- •For human resources compliance teams: Organize and reason over detailed employment-discrimination complaint events to support internal investigations and decision making.
Authors
Sriram Kannan, Swetha Saseendran, Vishnu Vardhan Reddy Kandi, Leslie Barrett, Madhavan Seshadri, Enrico Santus
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.