Relationship extraction from biomedical literature using Maximum Entropy based on rich features

Relation extraction is the detection and classification of interactions between named entities. Recently, a lot of studies have focused on the relation extraction from biomedical literatures. Although various approaches have been applied on this area, most methods are either pre-requires biomedical lexicons or parsing templates which are not suitable for complexities of biomedical literatures. In this paper, applying existing lexicons, we propose a Maximum Entropy method based on rich features to extract the interactions between the disease and treatment. This task is a multi-class biomedical relationship extraction. Experiment results show that the proposed method can achieve a comparable performance with the state-of-the-art generative biomedical relation extraction methods on multi-class disease-treatment interaction extraction.