TL;DR提出了一种简单而有效的依赖树导向的 LSTM-CRF 模型,以对命名实体识别(NER)任务中的完整依赖树进行编码并捕捉其相关属性,从而显著提高 NER 和实现领先水平,并发现依赖关系和依赖树提供的长距离交互是其主要原因。
Abstract
Dependency tree structures capture long-distance and syntactic relationships
between words in a sentence. The syntactic relations (e.g., nominal subject,
object) can potentially infer the existence of certain named entities. In
addition, the performance of a named entity recognizer cou