The biomedical abbreviation recognition and resolution (Barr) track: Benchmarking, evaluation and importance of abbreviation recognition systems applied to Spanish biomedical abstracts

Healthcare professionals are generating a substantial volume of clinical data in narrative form. As healthcare providers are confronted with serious time constraints, they frequently use telegraphic phrases, domain-specific abbreviations and shorthand notes. Efficient clinical text processing tools need to cope with the recognition and resolution of abbreviations, a task that has been extensively studied for English documents. Despite the outstanding number of clinical documents written worldwide in Spanish, only a marginal amount of studies has been published on this subject. In clinical texts, as opposed to the medical literature, abbreviations are generally used without their definitions or expanded forms. The aim of the first Biomedical Abbreviation Recognition and Resolution (BARR) track, posed at the IberEval 2017 evaluation campaign, was to assess and promote the development of systems for generating a sense inventory of medical abbreviations. The BARR track required the detection of mentions of abbreviations or short forms and their corresponding long forms or definitions from Spanish medical abstracts. For this track, the organizers provided the BARR medical document collection, the BARR corpus of manually annotated abstracts labelled by domain experts and the BARR-Markyt evaluation platform. A total of 7 teams submitted 25 runs for the two BARR subtasks: (a) the identification of mentions of abbreviations and their definitions and (b) the correct detection of short formlong form pairs. Here we describe the BARR track setting, the obtained results and the methodologies used by participating systems. The BARR task summary, corpus, resources and evaluation tool for testing systems beyond this campaign are available at: http://temu.inab.org.