Assessing top-$k$ preferences
This work addresses the challenge of more efficient and accurate evaluation for ranking systems, particularly benefiting researchers and practitioners in information retrieval, though it is incremental as it builds on existing preference judgment methods.
The paper tackles the problem of evaluating ranking systems using partial preference judgments instead of full ordering, introducing a new measure called 'compatibility' that better captures quality improvements in neural rankers, as demonstrated on the TREC 2019 dataset.
Assessors make preference judgments faster and more consistently than graded judgments. Preference judgments can also recognize distinctions between items that appear equivalent under graded judgments. Unfortunately, preference judgments can require more than linear effort to fully order a pool of items, and evaluation measures for preference judgments are not as well established as those for graded judgments, such as NDCG. In this paper, we explore the assessment process for partial preference judgments, with the aim of identifying and ordering the top items in the pool, rather than fully ordering the entire pool. To measure the performance of a ranker, we compare its output to this preferred ordering by applying a rank similarity measure.We demonstrate the practical feasibility of this approach by crowdsourcing partial preferences for the TREC 2019 Conversational Assistance Track, replacing NDCG with a new measure named "compatibility". This new measure has its most striking impact when comparing modern neural rankers, where it is able to recognize significant improvements in quality that would otherwise be missed by NDCG.