A Logical Approach to Criminal Case Investigation

Explainable AI can help address real challenges in criminal investigations: The paper identifies concrete problems forensic experts face—managing large datasets, analyzing small pieces of evidence in complex environments, and working within traditional laboratory structures—that can contribute to failed investigations and miscarriages of justice. XAI techniques that show their reasoning could help mitigate these issues.

Logical reasoning systems can systematically evaluate suspects: The research demonstrates that AI systems can be designed to evaluate suspects by systematically checking three key factors: motive, opportunity, and method. This structured approach mirrors established investigative principles and could help ensure investigations are more thorough and consistent.

Knowledge graphs and ontologies can organize investigative information: The study shows that creating formal knowledge structures (ontologies) that define types of motives and methods of murder, combined with case-specific knowledge graphs, allows AI systems to reason through complex criminal scenarios in a transparent way.

Transparency in AI reasoning is critical for justice system applications: By using logical approaches rather than "black box" AI, the paper emphasizes that criminal justice applications require explainability—investigators and courts need to understand why an AI system reached a conclusion, not just what conclusion it reached.

This approach is still in early stages: The application uses a fictional Sherlock Holmes story as a test case, indicating this is proof-of-concept research. Real-world implementation in actual criminal investigations would require significant additional development and validation.