English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Neural retrieval models are often trained on (subsets of) the millions ofqueries of the MS MARCO / ORCAS datasets and then tested on the 250 Robust04queries or other TREC benchmarks with often only 50 queries. In such setups,many of the few test queries can be very similar to queries from the hugetraining data -- in fact, 69% of the Robust04 queries have near-duplicates inMS MARCO / ORCAS. We investigate the impact of this unintended train-testleakage by training neural retrieval models on combinations of a fixed numberof MS MARCO / ORCAS queries that are highly similar to the actual test queriesand an increasing number of other queries. We find that leakage can improveeffectiveness and even change the ranking of systems. However, these effectsdiminish as the amount of leakage among all training instances decreases andthus becomes more realistic.

How Train-Test Leakage Affects Zero-shot Retrieval