10.5281/zenodo.1164161
https://zenodo.org/records/1164161
oai:zenodo.org:1164161
Koroleva Anna
Koroleva Anna
LIMSI-CNRS
Paroubek Patrick
Paroubek Patrick
LIMSI-CNRS
Automatic detection of inadequate claims in biomedical articles: first steps
Zenodo
2017
Inadequate Reporting
Spin
Biomedical Articles
Text Classification
Entity Extraction
2017-09-12
eng
10.5281/zenodo.1164160
https://zenodo.org/communities/miror
https://zenodo.org/communities/eu
Creative Commons Attribution 4.0 International
In this article we present the first steps in developing an NLP algorithm for automatic detection of inadequate reporting of research results (known as spin) in biomedical articles. Inadequate reporting consists in presenting the experimental treatment as having a greater beneficial effect than it was shown by the research results. We propose a scheme for an algorithm that would automatically identify important claims in the articles abstracts, extract possible
supporting information from the article and check the adequacy of the claims. We present the state of the art and our first experiments for three tasks related to spin detection: classification of articles according to the type of reported clinical trial; classification of sentences in the abstracts aimed at identifying mentions of the Results and Conclusions of the experiment; and extraction of some trial characteristics. For each task, we outline possible directions of further work.
European Commission
10.13039/501100000780
676207
Methods in Research on Research