ProUD: Probabilistic Ranking in Uncertain Databases

There are a lot of application domains, e.g. sensor databases, traffic management or recognition systems, where objects have to be compared based on vague and uncertain data. Feature databases with uncertain data require special methods for effective similarity search. In this paper, we propose an effective and efficient probabilistic similarity ranking algorithm that exploits the full information given by inexact object representations. Thereby, we assume that the objects are given in form of discrete probabilistic object locations in particular several object snapshots with confidence values. Based on the given object representations, we suggest diverse variants of probabilistic ranking schemes. In a detailed experimental evaluation, we demonstrate the benefits of our probabilistic ranking approaches. The experiments show that we can achieve high quality query results while keeping the computational cost quite small.