Unsupervised linear score normalization revisited

We give a fresh look into score normalization for merging result-lists, isolating the problem from other components. We focus on three of the simplest, practical, and widelyused linear methods which do not require any training data, i.e. MinMax, Sum, and Z-Score. We provide theoretical arguments on why and when the methods work, and evaluate them experimentally. We find that MinMax is the most robust under many circumstances, and that Sum is— in contrast to previous literature—the worst. Based on the insights gained, we propose another three simple methods which work as good or better than the baselines. Categories and Subject Descriptors H.3.3 [Information Storage and Retrieval]: Information Search and Retrieval Keywords Score Normalization, Distributed Retrieval

• Type: Journal
• Year: 2012
• Where: SIGIR
• Authors: Ilya Markov, Avi Arampatzis, Fabio Crestani
• Keywords: Information Storage And Retrieval, Information Technology, Linear Methods, SIGIR 2012, Z Score