Use of Bayesian networks in forensic soil casework.

Area of Science:

• Forensic Science
• Computational Science

Background:

• Forensic soil comparisons offer significant evidential value but present complexities due to multiple analytical methods and influencing factors.
• Bayesian networks offer a robust framework for managing and interpreting complex probabilistic information in forensic casework.

Purpose of the Study:

• To explore the application of Bayesian networks in forensic soil comparison.
• To provide a structural template for constructing case-specific Bayesian networks.
• To demonstrate the utility of Bayesian networks for evaluating source-level and activity-level propositions in forensic soil analysis.

Main Methods:

• The study outlines the structural components of a Bayesian network relevant to forensic soil analysis.
• It details the types of input and output data required for network construction and evaluation.
• Two case examples are presented: one focusing on source-level propositions and the other on activity-level propositions.

Main Results:

• The developed Bayesian network structures provide a clear framework for integrating diverse soil evidence data.
• The evaluated examples demonstrate how to apply these networks to specific forensic scenarios.
• Sensitivity analysis of the networks can identify key factors influencing the evidential weight of soil comparisons.

Conclusions:

• Bayesian networks are a valuable tool for forensic practitioners dealing with complex soil comparison cases.
• The provided templates and examples facilitate the construction and application of case-specific networks.
• This approach aids in assessing the evidential value of soil data and highlights areas for future research in forensic soil science.