基于Bert模型的框架语义角色标注方法

Bert-Based Frame-Semantic Role Labeling Method

语义角色标注是自然语言处理领域的一个研究热点,其成果已经被广泛应用于智能问答、信息抽取、知识图谱等领域。提出了一种基于Bert模型的框架语义角色标注方法,将序列标注的思想应用到框架语义角色标注中。由于框架元素种类过多,构建了框架元素到全局语义角色的映射关系。在Bert预训练网络的基础上,增加双向LSTM层以及CRF层,构建序列标注模型,该模型同时考虑上下文信息、词元信息以及框架类型信息,在FrameNet语义角色标注数据上的测试性能优于对照模型,证明了该方法的有效性。

Semantic role labeling is one of the hot topics of natural language processing.It has been widely used in question answering,information extraction,knowledge graph,etc.This paper presents a frame-semantic role labeling method based on the famous pre-trained model named Bert.Our method adopts the idea of sequence labeling to label frame-semantic roles.As there are too many frame elements,we create the mappings between frame elements and global semantic roles.Our sequence labeling model is constructed on Bert network with a bidirectional LSTM layer and a CRF layer.The model takes into consideration of contexts,lexical units and frame types and outperforms the control models,which proves the effectiveness of our method.