ExB Medical Text Miner

We present ExB Medical Text Miner – a text mining pipeline for processing biomedical documents. This application employs stateof-the-art Named Entity Recognition, using linguistic features and word embeddings in a fully-connected second-order Conditional Random Field model, as well as a novel two-stage Relation Extraction module that first detects entity-level relations using a Support Vector Classifier, then identifies document-level relations by measuring their relevance according to a document topic classification model.