Automatic Drug-Drug Interaction Detection: A Machine Learning Approach With Maximal Frequent Sequence Extraction

Abstract

[EN] A Drug-Drug Interaction (DDI) occurs when the effects of a drug are modified by the presence of other drugs. DDIExtraction2011 proposes a first challenge task, Drug-Drug Interaction Extraction, to compare different techniques for DDI extraction and to set a benchmark that will enable future systems to be tested. The goal of the competition is for every pair of drugs in a sentence, decide whether an interaction is being described or not. We built a system based on machine learning based on bag of words and pattern extraction. Bag of words and other drug-level and character-level have been proven to have a high discriminative power for detecting DDI, while pattern extraction provided a moderated improvement indicating a good line for further research.