We investigate to what extent people making relevance judgments for a reusable IR test collection are exchangeable. We consider three classes of judge: gold standard judges, who are topic origi- nators and are experts in a particular information seeking task; sil- ver standard judges, who are task experts but did not create topics; and bronze standard judges, who are those who did not define topics and are not experts in the task. Analysis shows low levels of agreement in relevance judgments between these three groups. We report on experiments to determine if this is sufficient to invalidate the use of a test collection for mea- suring system performance when relevance assessments have been created by silver standard or bronze standard judges. We find that both system scores and system rankings are subject to consistent but small differences across the three assessment sets. It appears that test collections are somewhat robust to changes of judge even if these judges vary widely in task and topic expertise. Bronze stan- dard judges may be able to substitute for topic and task experts, with some caution regarding relative system performance, but gold standard judges are preferred.
Proceedings Title: Proceedings of the 31th Annual International ACM SIGIR Conference on Research and Development Information Retrieval
Conference Dates: July 21-25, 2008
Conference Location: Singapore, SN
Conference Title: ACM SIGIR 2008 (Special Interest Group for Information Retrieval)
Pub Type: Conferences
enterprise search, information retrieval, relevance assessment, test collections