Multi-class Relationship Extraction from Biomedical Literature Using Maximum Entropy

Relation extraction is a challenging task in biomedical text mining due to the complex of sentences in the biomedical literature. In this paper, we address multi-class relationship extraction problem from biomedical literature using Maximum Entropy model with simple word features. The proposed method is applied to extract the protein-protein interactions. Experiments show the method achieves an accuracy of 73.4% in the corpora built based on the HIV-1 Human Protein Interaction Database, which is a promising result compare to previous works.