Rank-Biased Precision Reloaded: Reproducibility and GeneralizationAdvances in Information Retrieval

In this work we reproduce the experiments presented in the paper entitled “Rank-Biased Precision for Measurement of Retrieval Effectiveness”. This paper introduced a new effectiveness measure – Rank- Biased Precision (RBP) – which has become a reference point in the IR experimental evaluation panorama. We will show that the experiments presented in the original RBP paper are repeatable and we discuss points of strength and limitations of the approach taken by the authors. We also present a generalization of the results by adopting four experimental collections and different analysis methodologies.