Comparing metrics across TREC and NTCIR: : the robustness to pool depth bias

Abstract

Test collections are growing larger, and relevance data constructed through pooling are suspected of becoming more and more incomplete and biased [1]. Relevance data are incomplete if there exist some relevant documents among the unjudged documents in the test collection. Furthermore, incomplete relevance data are biased if they represent some limited… (More)
DOI: 10.1145/1390334.1390454

Topics

2 Figures and Tables

Cite this paper

@inproceedings{Sakai2008ComparingMA, title={Comparing metrics across TREC and NTCIR: : the robustness to pool depth bias}, author={Tetsuya Sakai}, booktitle={SIGIR}, year={2008} }