An Extended CLEF eHealth Test Collection for Cross-lingual Information Retrieval in the Medical Domain
Abstract
We present a test collection for medical cross-lingual information retrieval. It is built on resources used by the CLEF eHealth Evaluation Lab 2013-2015 in the patient-centered information retrieval tasks and greatly improves applicability and reusability of the official data.
The document set is identical to the official one used for the task in 2015 and contains about 1 million English medical webpages. The query set consists of 166 queries used during the three years of the campaign as test queries, now available in eight languages (the original English and human translations into Czech, French, German, Hungarian, Polish, Spanish and Swedish).
The extended test collection provides extensively improved relevance judgements which almost doubled the amount of the officially assessed query-document pairs. This paper describes the content of the extended version of the test collection, details of query translation and relevance assessment, and state-of-the-art results on this collection.
The dataset can