【摘要】 挖掘新兴产业中的检测信息有利于发现检测机构的检测能力，促进机构合作，加速产业升级。针对新兴产业检验检测数据存在语义混乱、一个主实体对应多个客实体的问题，本文提出一种混合关系标签的深度神经网络实体关系抽取模型。在输入层，构建基于关系的标签，并与语义信息拼接形成模型的输入，增强了模型对不同关系的区分度；在特征提取层，使用双向长短期神经网络与卷积神经网络，从整体与局部提升模型对主客实体特征的抽取能力，同时引入注意力机制，削弱无关特征的影响，提升模型对主客实体的识别能力。实验结果表明，该模型不仅能有效识别出新兴产业检验检测领域的实体，而且能精准判断实体之间的关系，取得了较好的结果。更多还原

【Abstract】 Mining detection information in emerging industries is conducive to discovering detection capabilities of detection institutions, promoting cooperation between institutions and accelerating industrial upgrading. In order to solve the problem of semantic confusion in the inspection data of emerging industries, a deep neural network entity relation extraction model based on mixed relation labels is proposed in this paper. At the input layer, the labels based on relationships are constructed and the input of the model is spliced together with semantic information to enhance the differentiation of different relationships. At the feature extraction layer, bidirectional long and short-term neural network and convolutional neural network are used to improve the feature extraction ability of the model. Meanwhile, attention mechanism is introduced to weaken the influence of irrelevant features and improve the recognition ability of the model to host and object entities. The experimental results show that the model can not only identify the entities in the inspection and testing field of emerging industries, but also accurately judge the relationship between entities, achieving good experimental results.更多还原