Decision-theoretic file carving

This article explores a novel approach to file carving by viewing it as a decision problem. This allows us to design algorithms that produce best-effort results under given resource constraints. Resource-constrained carving is important for digital forensic triage, as well as for e-discovery, where a reduction in carving time may be preferred to completeness. In this work we give a formal definition of decision-theoretic file carving. As an illustration, we developed a JPEG file carving tool using the described decision-theoretic algorithm. We then examine the results of decision-theoretic file carving compared with linear carving methods to demonstrate when decision-theoretic carving is most useful.