The algorithm ranks candidates on the basis of fingerprint descriptors and three probability factors. Computer disk files were created to capture data relating to (1) the age frequency distribution of persons arrested in New York State for each of 12 penal law charge categories by sex and subdivision of the State (upstate or New York City), (2) the probability of rearrest on any of the 12 charges after an arrest on a first charge within 10 years of first arrest, and (3) the probability of a male or female between 16 and 65 years being arrested on any of the 12 charges. The algorithm is a linear combination of elements of these files and a factor derived from sums of differences of latent and file print ridge counts. A sample was selected and programs written to test the ranking procedure. The test results are presented and incorporated into a cost model for the Division of Criminal Justice Services Special Services' Unit. Based on the evaluation, it is recommended that the system be implemented and that an indepth exploration of modifications to the existing arrest fingerprint system be conducted to enhance throughout time. Details of the system's operation along with data produced are appended. (Author abstract modified)
Latent Fingerprint Identification System Report
NCJ Number
81370
Date Published
1981
Length
107 pages
Annotation
The candidate-ranking algorithm developed for use in the New York State latent fingerprint identification system is described, and its use is evaluated, followed by an analysis of the development of similar systems by other criminal justice agencies.
Abstract