Joint Biomedical Entity and Relation Extraction Based on Feature Filter Table Labeling

Zhaojie Sun, Linlin Xing, Longbo Zhang, Hongzhen Cai, Maozu Guo

Abstract: Joint biomedical entity and relation extraction is essential in biomedical text mining. It automatically identifies entities and uncovers the relation between them from biomedical texts. However, due to the relatively complex semantics of biomedical texts, the current methods are unable to effectively leverage the interaction between the two subtasks. In this work, in order to use the interaction between the subtasks, we propose to model entity labels and relation labels with table-filling. We assume that the table structure facilitates the information exchange between entities and relations. Additionally, a feature filtering module is designed in the model to enhance this interaction. After passing through the feature filtering module, the table was constructed based on the selected global features. Our model was evaluated on two tasks, the task of extracting adverse drug events between drug and disease entities, and the task of extracting interaction between drug entities. Compared with the state-of-the-art systems in these tasks, our model improved the F1 scores of the first task by 0.97% in entity recognition and 1.43% in relation extraction, and that of the second task by 1.14% in relation extraction.