Building Deep Learning Models for Evidence Classification from the Open Access Biomedical Literature
This repository describes a Research Object for the data and analysis reported in the paper:
Burns, et al (2019) "Building Deep Learning Models for Evidence Classification from the Open Access Biomedical Literature, submitted to Biocuration 2019.
The manifest.ttl file provides linked data for the resources (software and data) used in the data processing that generated this paper. This involves links to three datasets hosted on Zenodo and three software tools hosted on Github. None of these resources are currently linked data / research objects, but we will seek to incorporate more detailed semantic representations of these elements going forward.
- https://github.com/SciKnowEngine/evidX/releases/tag/v0.1.0 - Simple TensorFlow Classifiers used to classify text of subfigure captions by method type.
- https://github.com/SciKnowEngine/UimaBioC - Text preprocessing pipelines
- https://github.com/SciKnowEngine/lapdftext - PDF image and text extraction tools
- https://doi.org/10.5281/zenodo.1315036 - 'Molecular Biology Open Access Pubmed Word and Sentence Representations'
- https://doi.org/10.5281/zenodo.1315021 - 'Method Classification of Open Access INTACT Molecular Interaction data.'