•  
  •  
 

Article Type

Research Article

Abstract

Witness testimony is an important source of evidence in criminal proceedings, but it may contain contradictions, incomplete details, or conflicts with other case-file materials. This paper proposes a neutrosophic event-graph legal AI framework for detecting materially contested claims in witness testimonies under the Egyptian criminal-procedure context. The framework converts testimony and related records into structured claims containing actor, action, object, time, location, source, and modality. These claims are then connected through an event graph and evaluated using neutrosophic components of support, indeterminacy, and opposition. The system produces source-grounded legal-review alerts when a claim has sufficient opposition from other claims or records. It does not decide whether a witness is truthful, does not determine false testimony, and does not make findings of guilt or innocence. A controlled synthetic case study demonstrates how the framework identifies time-location and source-based contradictions while preserving uncertainty. The paper also presents robustness analysis, diagnostic comparison with controlled baselines, legal safeguards, and a practical implementation model for a secure legal-review system or mobile application. The proposed framework supports legal professionals by organizing evidential conflict, improving traceability, and prioritizing claims that require human review.

Keywords

Neutrosophic logic, Legal AI, Witness testimony, Contradiction detection, Event graph, Egyptian criminal procedure, Legal-review alerts, Evidential reasoning, Source-grounded explanation, Legal decision support

Download

Share

COinS