This report describes the process of developing a sanitized close non-match database, called the International Close Non-Match Library (ICNML), for research and training to be shared with trusted members of the law enforcement community.
The authors report on the development of a sanitized close non-match (CNM) database for research and training to be shared with trusted members of the law enforcement agency (LEA) community. CNMs are prints from different sources that show high levels of agreement, or correspondence, that could mislead an examiner to erroneously conclude that the different marks were from the same source. The database is called the International Close Non-Match Library (ICNML) and consists of collected known ground-truth marks and prints from 100 donors and searched marks through multiple Automated Fingerprint Identification Systems (AFIS) databases, both in the US and internationally, to obtain CNMs. One research question the authors explored was whether specific groups of criteria, or red flags, could be used to retrieve a higher number of CNMs when they were searched in large AFIS databases. If they did, those criteria could be developed into training for latent print examiners (LPEs) on when to exercise additional caution around identification decisions that may be more likely to indicate a CNM.