In addition, the inter-sentential relations account for approximately 1/3 of all relations, signifying that traditional sentence-level relation extraction methods may not be appropriate to get satisfactory results. According to the document-level annotation, it is hard to tell which sentence(s) convey(s) the meaning of a specific relation, since an entity can be mentioned multiple times in different sentences in an abstract and the offsets of related entities, which can be used to identify the unique mention of an entity in an abstract, are not given. 1, given an abstract, entity mentions with entity offsets and related entity pairs are annotated. without giving the specific sentence that conveys a relation). Semeval-2010 Task 8 ), the CID relations are annotated only at the document level (i.e. In this chemical disease relation (CDR) corpus, different from traditional sentence-level relation classification tasks (e.g. 1 shows an example of an abstract from the challenge corpus with its annotations. However, most of the existing relation extraction methods focus merely on intra-sentential relations, which is apparently insufficient in capturing inter-sentential ones.Ī benchmark document-level relation extraction task was proposed in the BioCreative V challenge, in which participating systems were asked to return all possible chemical-disease (CD) pairs that express document-level chemical-induced disease (CID) relations in a given abstract. ”, the inter-sentential relation “ fusidic acid” - induced “ nausea” can be obtained only by integrating the semantic information in both sentences. There were no serious clinical side effects, but dose reduction was required in two patients because of nausea. For example, in the paragraph “ Five of 8 patients improved during fusidic acid treatment: 3 at two weeks and 2 after four weeks. Inter-sentential relations can account for a substantial proportion and convey important meanings, especially in the biomedical domain. Semantic relations in texts can be expressed either intra-sententially (within a sentence) or inter-sententially (cross sentence boundaries). Our work can facilitate the research on document-level biomedical text mining. Sequence labeling method can be successfully used to extract document-level relations, especially for boosting the performance on inter-sentential relation extraction. Also, our method achieved an F1-score of 85.1% on n2c2-ADE sub-dataset. Our proposed method obtained an F1-score of 63.5% on BioCreative V chemical disease relation corpus, and an F1-score of 54.4% on inter-sentential relations, which was 10.5% better than the document-level classification baseline. Besides, the sequence labeling framework enables Bio-Seq to take advantage of the interactions between relations, and thus, further improves the precision of document-level relation extraction. In the method, sequence labeling framework is extended by multiple specified feature extractors so as to facilitate the feature extractions at different levels, especially at the inter-sentential level. We propose a novel sequence labeling-based biomedical relation extraction method named Bio-Seq. However, most existing methods either focus on extracting intra-sentential relations and ignore inter-sentential ones or fail to extract inter-sentential relations accurately and regard the instances containing entity relations as being independent, which neglects the interactions between relations. Both intra- and inter-sentential semantic relations in biomedical texts provide valuable information for biomedical research.