UNSTRUCTURED: Background: Automatically extracting relations between chemicals and diseases plays an important role in biomedical text mining. Chemical-disease relation (CDR) extraction aims at extracting complex semantic relationships between entities in documents, which contain intra- and inter-sentence relations. Most previous methods do not consider dependency syntactic information across the sentences, which are very valuable for the relations extraction task, in particular for extracting the inter-sentence relations accurately. Methods: In this paper, we propose a novel end-to-end neural network based on the graph convolutional network (GCN) and multi-head attention. To improve the performance of inter-sentence relation extraction, we construct the document-level dependency graph to capture the dependency syntactic information across sentences. GCN is applied to capture the feature representation of the document-level dependency graph. The multi-head attention mechanism is employed to learn the relative important context features from different semantic subspaces. To enhance the input representation, the deep context representation is used in our model instead of traditional word embedding. Results: The experimental results show that our method achieves an F-score of 63.5% which is superior to other state-of-the-art methods. The GCN model can effectively exploit the across sentence dependency information to improve the performance of inter-sentence CDR extraction. Both the deep context representation and multi-head attention are helpful in CDR extraction task.
UNSTRUCTURED: Background: Automatically extracting relations between chemicals and diseases plays an important role in biomedical text mining. Chemical-disease relation (CDR) extraction aims at extracting complex semantic relationships between entities in documents, which contain intra- and inter-sentence relations. Most previous methods do not consider dependency syntactic information across the sentences, which are very valuable for the relations extraction task, in particular for extracting the inter-sentence relations accurately. Methods: In this paper, we propose a novel end-to-end neural network based on the graph convolutional network (GCN) and multi-head attention. To improve the performance of inter-sentence relation extraction, we construct the document-level dependency graph to capture the dependency syntactic information across sentences. GCN is applied to capture the feature representation of the document-level dependency graph. The multi-head attention mechanism is employed to learn the relative important context features from different semantic subspaces. To enhance the input representation, the deep context representation is used in our model instead of traditional word embedding. Results: The experimental results show that our method achieves an F-score of 63.5% which is superior to other state-of-the-art methods. The GCN model can effectively exploit the across sentence dependency information to improve the performance of inter-sentence CDR extraction. Both the deep context representation and multi-head attention are helpful in CDR extraction task.