A computational ecosystem to support eHealth Knowledge Discovery technologies in Spanish

Abstract

Elsevier has created a COVID-19 resource centre with free information in English and Mandarin on the novel coronavirus COVID-19. The COVID-19 resource centre is hosted on Elsevier Connect, the company's public news and information website. Elsevier hereby grants permission to make all its COVID-19-related research that is available on the COVID-19 resource centre-including this research content-immediately available in PubMed Central and other publicly funded repositories, such as the WHO COVID database with rights for unrestricted research re-use and analyses in any form or by any means with acknowledgement of the original source. These permissions are granted for free by Elsevier for as long as the COVID-19 resource centre remains active.

Automatic extension of corpora from the intelligent ensembling of eHealth knowledge discovery systems outputs

“…Using an optimization search algorithm, the space of possible ensemblers that can be built with the given strategies is explored to find the optimal configuration for a concrete corpus. The eHealth-KD v2 corpus [22] was taken as a case of study, resulting in the automatic annotation of the relevant key phrases and relations present across 8 000 sentences. The submissions presented…”

“…Many information extraction competitions (e.g., eHealth-KD 1 , IberLegal 2 , and The eHealth-KD 2019 (or eHealth-KD v2 [21]) corpus, is a noteworthy example of information extraction related corpora available in the literature [22]. It is made of three types of annotations: entities, relations, and attributes, all of them manually annotated in a collection of Spanish sentences from the medical domain.…”

“…For reference purposes, the human baseline presented by Piad et al [22] is included in Table 5. The performance of the human baseline approximates the best performance that can be achieved by any automatic annotation system.…”

“…The collection of 15 seare part of the original annotation scheme of the eHealth-KD v2 corpus, they were not considered in the context of the challenge and, therefore, are not available for the automatic construction of this ensembled corpus. Just as the original corpus, the most represented entity classes are Concept and Action[22]. The most represented relations are target, subject, and in-context.…”

“…compares in the three evaluation scenarios the performance achieved by the selected ensembler against the baseline and the best performing submission to the challenge (TALP-UPC). Additionally, the human baseline presented by Piad et al[22] is included. The ensembler was able to outperform TALP-UPC's performance in all scenarios but the third one.…”