Extraction of the relevant population from a forensic database.

Area of Science:

• Forensic Science
• Statistics
• Database Management

Background:

• Bayesian statistical evaluation of forensic evidence necessitates hypothesis formulation.
• Defence hypotheses often involve attributing traces to arbitrary members of a relevant population.
• The composition of the relevant population varies per case, challenging fixed database use.

Purpose of the Study:

• To introduce a methodology for filtering forensic databases.
• To select relevant items for statistical evaluation based on case-specific hypotheses.
• To enhance the accuracy of forensic evidence assessment.

Main Methods:

• Development of a database filtering methodology.
• Application to six scenarios including fibre, textile, and glass evidence.
• Utilizing SQL code for extracting relevant population items from databases.
• Visual feedback through item images for examiner review.

Main Results:

• A procedure for filtering database contents to identify relevant items is presented.
• SQL code demonstrates efficient extraction of items representing the defined relevant population.
• Visualizations aid in identifying and correcting erroneous codes or side effects.

Conclusions:

• The proposed filtering procedure is effective for forensic database management.
• Accurate demarcation of the relevant population is crucial for the procedure's success.
• This method improves the statistical evaluation of evidential value in forensic science.