There is an increasing interest, both on the academic side as well as for industry, to facilitate more efficient access to information on chemical compounds and drugs (chemical entities) described in repositories of unstructured data, including scientific articles, patents or health agency reports. In order to achieve this goal, a crucial aspect is to be able to identify mentions of chemical compounds automatically within text as well as to index whole documents with the compounds described in them. The recognition of chemical entities is also crucial for other subsequent text processing strategies, such as detection of drug-protein interactions, adverse effects of chemical compounds and their associations to toxicological endpoints or the extraction of pathway and metabolic reaction relations.Despite its importance, only a very limited number of publicly accessible chemical compound recognition systems have been released. In contrast to this, a considerable number of methods and strategies to recognize chemicals in text have been proposed. One of the main bottlenecks currently encountered to implement and compare the performance of such systems is the (a) lack of suitable training/test data, (b) the intrinsic difficulty in defining annotation guidelines of what actually constitutes a chemical compound or drug, (c) heterogeneity in terms of scope and textual data sources used, as well as (d) limited evaluation efforts carried out so far.A total of 27 teams submitted results for the proposed CHEMDNER task. Teams were provided with the manual annotations of 7,000 abstracts to implement and train their systems and then had to return predictions for the 3,000 test set abstracts during a short period of time. When directly comparing the automated results against the manually labeled Gold Standard annotations, the best team reached an F-score (the harmonic mean between precision of coverage) of 87% on the Chemical Entity Mention (CEM) task (http://www.biocreative.org/tasks/biocreative-iv/chemdner)

CHEMDNER task: Automatic recognition of chemical entities in text

248th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.