Beyond C.S.I.: The Rise of Computational Forensics

Recommendations from the National Academy of Sciences

Illustration: Emiliano Ponzi

Recognizing the myriad problems with classical forensics techniques, the U.S. Congress directed the National Academy of Sciences in late 2005 to look into the matter. A committee of legal experts, practitioners, and scientists (including the author) was formed, and after a two-year study, which included testimony from experts in a variety of forensic disciplines, the committee released its report in February 2009.

The 13 wide-ranging recommendations included four areas where computational forensics could be useful:

Recommendation 1 calls for the establishment of a new federal entity, the National Institute of Forensic Sciences, to focus on, among other things, competitive peer-reviewed research, the generation of accurate data from the forensic science community, strong support for forensic methodologies and practices, the assessment of new technologies in forensic investigations, and the use of standardized terminology and reporting. Clearly, computational methods are new technologies that hold promise.

Recommendation 3 calls for studies to examine the validity of forensic methods and to develop quanti¿able measures of reliability and accuracy of forensic techniques. Computational methods can allow large-scale testing to accomplish this objective.

Recommendation 5 concerns the need to conduct research on human observer bias as well as human error in forensic examinations. Computational methods can help eliminate bias and minimize error.

Recommendation 12 deals exclusively with fingerprints, including the need for baseline standards to be used with computer algorithms to map, record, and recognize features in fingerprint images. Also recommended is that a research agenda be developed for the continued improvement, refinement, and characterization of the accuracy of these algorithms (including quantification of error rates). —Sargur N. Srihari