SLR Recommendation - CapeStart

PythonAWS

Pharma NLP is a service offered by CapeStart to make sense of big text data, turn transcriptions into value, improve pharmacovigilance and acquiring new targets in the biomedical field.

Systematic Literature Review (SLR) is a systematic means of collecting, critically assessing, integrating, and presenting findings from across various research papers on a research issue or specialized topic of interest. This project focuses on recommending a researcher the most relevant SLR that he is seeking out for when a user sample is given. The recommendation is given out on an actual context level.

I was working in tandem with my team colleague Joseph on this project. Some of the important development involved in the project are:

  • Data collection from pubmed
  • Annotation of data
  • Vectorization of the data
  • Feature Engineering that includes PICO extraction on EBM based bio medical data
  • Model development that makes use of Bert model with additional tuning and highest recall
  • Parallel model training using AWS gpus
  • Ranking mechanism to output recommendations to user