An automated method to extract information in the biomedical literature about interactions between drugs

Text mining techniques are useful in extracting hidden information about the biomedical interactions such as Protein-Protein, Drug-Drug and Protein-Drug. Recently, there is an increased interest in automated methods due to the vast growth in the volume of published text regarding biomedical interactions. This work mainly focuses on extraction of Drug-Drug interactions (DDIs) in biomedical research articles from well-known databases such as DrugBank and MedLine. The proposed approach is developed based on feature engineering through natural language processing (NLP) techniques such as bag-of-words approach, tokenization, part-of-speech (POS) tagging, lemmatization and so on. This uncomplicated and easy to implement set of features are combined into a feature vector which is used to train a machine learning model. The effectiveness of the proposed approach was measured by conducting several experiments on the “DDI Extraction 2013” corpus. The system showed encouraging F-measure value of 76.9%.