基于序列标注的事件联合抽取方法

事件抽取是自然语言处理领域的重要研究方向。传统的事件类型抽取系统采用分类方式，无法解决跨句子的事件角色和事件类型匹配问题。为了解决该问题，提出一种基于序列标注的事件联合抽取模型，结合卷积神经网络(convolutional neural networks, CNN)与长短期记忆网络(long short-term memory, LSTM)提取全局特征和局部特征；并在浅层LSTM层共享参数实现联合抽取，以序列标注方式抽取事件论元并匹配事件类型。实验结果表明，模型能有效提取司法领域的文档事件信息。

Joint event extraction based on sequence annotation

Event extraction is an important research in the field of natural language processing. In traditional event extraction systems, event types as a classification task cannot be matched with event arguments. In this paper, a joint event extraction model based on sequence annotation is proposed to solve event matching across sentences. Global and local features are extracted by convolutional neural networks (CNN) and long short-term memory (LSTM). The joint extraction is realized by sharing parameters in the shallow LSTM layer, event types and arguments are extracted and matched by sequence annotation. The experimental results in the judicial field corpus show the effectiveness of proposed model.