The construction of medical knowledge graph based on electronic medical records is of great significance to the development of medical technology,where entity and relation extraction plays a pivotal role.In this paper,we propose a Parallel Interactive Attention Network(PIAN)which can fully exploit the correlation between entity and relation.Since there are few Chinese medical entity and relation annotation datasets,we construct an entity and rela-tion extraction dataset based on Chinese electronic medical records(CEMRIE),formulate the corpus annotation specification with medical experts,and give the benchmark results based on our proposed model.
关键词
实体关系联合抽取/双向特征交互模块/自注意力机制/中文电子病历/数据集标注与构建
Key words
joint entity and relation extraction/bidirectional feature interaction module/self-attention mechanism/chi-nese electronic medical record/dataset annotation and construction
维普
万方数据