BACKGROUND
With the rapid expansion of biomedical literature, biomedical information extraction (IE) has attracted more and more attention by researchers, especially the relation extraction (RE) between two entities is a long-term research topic.
OBJECTIVE
This paper focuses on two multi-class relation extraction tasks of BioNLP 2019 Open Shared Tasks: relation extraction of Bacteria-Biotope task (BB-rel) and binary relation extraction of plant seed development task (SeeDev-binary). In essence, these two tasks are aimed to extract the relation between annotated entity pairs from biomedical texts, which is a challenging problem.
METHODS
The traditional research methods adopted feature-based or kernel-based methods, and achieved good performance. For these tasks, we propose a deep learning model based on a combination of several distributed features, such as domain-specific word embedding, part-of-speech (POS) embedding, entity type embedding, distance embedding and position embedding. The Multi-Head attention mechanism is used to extract the global semantic features of a whole sentence. Meanwhile, we introduce dependency type feature and shortest dependency path connecting two candidate entities in the syntactic dependency graph to enrich the feature representation.
RESULTS
Experiments show that our proposed model has excellent performance in biomedical relation extraction, achieves the F1-scores of 65.56% and 38.04% on the test sets of these two tasks, respectively. Especially in SeeDev-binary task, the F1-score of our model is superior to other existing models and achieves state-of-the-art performance.
CONCLUSIONS
We demonstrated that the Multi-Head attention mechanism can learn relevant syntactic and semantic features in different representation subspaces and different positions to extract comprehensive feature representation. Moreover, syntactic dependency features can improve the performance of the model by learning dependency relation between the entities in biomedical texts.