Ricciardi et al 2020 Ceres Main Results

This dataset was created for the Ceres2030 Team 6, who conducted a semi-automatic systematic review of over 18,000 articles. The goal was to identify solutions for small farms to adapt to climate change - we focused on water scarce environments. This analysis is under review at the academic journal Nature Sustainability. A link to the article will be provided once published. The manuscript's current citation is:<br>Ricciardi, V., Wane, A., Singh Sidhu, B., Goode, C., Solomon, D., McCullough, E., Diekmann, F., Porciello, J., Jain, M., Randall, N., Mehrabi, Z.Evidence synthesis: Funding research for small-scale farmers in water scarce regions to achieve SDG 2.3. In Review. Nature Sustainability.<br>We use manually coded abstracts to build a natural language processing (NLP) model to include or exclude abstracts to our study. A deep learning model was developed using Bidirectional Encoder Representations from Transformers (BERT) and a cross entropy function to classify abstracts. This model performed better than other classifiers (e.g., optimized support vector machine (SVM) and naive bayes) across accuracy, precision, and recall metrics. This dataset is the predicted results.<br>This dataset is an input file for use in:https://github.com/vinnyricciardi/Ricciardi_etal_2020_ceres/blob/master/full_text_analysis.R<br><br>The data filename referred to in the code is: Validation Survey_February 17, 2020_12.13.csv<br>