Kelly L, Sachan S, Ni L, Almaghrabi F, Allmendinger R, Chen YW. Explainable artificial intelligence for digital forensics: opportunities, challenges and a drug testing case study. In: Suresh Shetty B, Shetty P, editors. Digital forensic science. London: UK. Digital Forensic Science by IntechOpen; 2020. p.424. doi: 10.5772/intechopen.93310

Forensic analysis is typically a complex and time-consuming process requiring forensic investigators to collect and analyse different pieces of evidence to arrive at a solid recommendation. Our interest lies in forensic drug testing, where evidence comprises a multitude of experimentally obtained data from samples (e.g. hair or nails), occasionally combined with questionnaire data, with a goal of quantifying the likelihood of drug use. The availability of intelligent data-driven technologies can support holistic decision-making in such scenarios, but this needs to be done in a transparent fashion (as opposed to using black-box models). To this end, this book chapter investigates the opportunities and challenges of developing interactive and explainable Artificial Intelligence (XAI) systems to support digital forensics and automate the decision-making process to enable fast and reliable generation of evidence for the court of law. Relevant XAI techniques and their applications in forensic testing, including feature section, missing data handling, XAI for multi-criteria and interactive learning, are discussed in detail. A case study on a forensic science company is used to demonstrate the real challenges of forensic reporting and potential for making use of forensic data to pave the way for future research towards XAI-driven digital forensics.

Share on: