Abstract
Abstract. In this paper, we aim to present a novel application domain for recommender systems: police photo lineups.
Photo lineups play a significant role in the eyewitness identification prosecution and subsequent conviction of suspects. Unfortunately, there are many cases where lineups have led to the conviction of an innocent persons.
One of the key factors contributing to the incorrect identification is unfairly assembled (biased) lineups, i.e. that the suspect differs significantly from all other candidates. Although the process of assembling fair lineup is both highly important and time-consuming, only a handful of tools are available to simplify the task.
We describe our work towards using recommender systems for the photo lineup assembling task. Initially, two non-personalized recommending methods were evaluated: one based on the visual descriptors of persons and the other their content-based attributes.
Next, some personalized hybrid techniques combining both methods based on the feedback from forensic technicians were evaluated. Some of the personalized techniques significantly improved the results of both non-personalized techniques w.r.t. nDCG and recall@top-k.