SemEval-2013 Task 9: Extraction of Drug-Drug Interactions from BioMedical Texts

Project done for the AHLT (Advanced Human Languages Technologies) course (Master in Artificial Intelligence at UPC)

Authors:

• Karen Lliguin
• Albert Rial

This repository contains different approaches to address the SemEval 2013 Task 9. This task concerns the recognition of drugs and extraction of drug-drug interactions that appear in biomedical literature. It is divided in two subtasks:

• Task 9.1: Recognition and classification of drug names.
• Task 9.2: Detection and classification of drug-drug interactions between pairs of drugs.

In the source folder you will find several jupyter notebooks containing each one a different approach. There are also two notebooks (one per subtask) called report.ipynb where you can find the explanation of each approach taken and the results obtained.

Results

The best results obtained are the following:

Task 9.1 (NERC)

• Devel set: precision 0.95, recall 0.73, F1 0.79
• Test set: precision 0.9, recall 0.64, F1 0.69

Task 9.2 (DDI)

• Devel set: precision 0.68, recall 0.62, F1 0.65
• Test set: precision 0.52, recall 0.59, F1 0.55