Mining medical academic articles using recurrent neural networks

Abstract

In this thesis, we present our methods and results for mining the MedMentions data [Mohan and Li, 2019]. We propose a pipeline for combining mention classification and mention disambiguation. We will use the Long Short Term Memory (LSTM) neural network architecture for mention detection and mention classification. Also we use nearest neighbour search using embeddings in the Unified Medical Language System (UMLS) concepts in order to disambiguate mentions.