Introduction
Challenge 1:常见词、歧义词和代词在事件中的频繁使用使它们更难被发现
Generality – taken home <Transport>
Ambiguty 1 – campaign in Iraq <Attack>
Ambiguty 2 – political campaign <Elect>
Coreference – Either its bad or good <Marry>
Challenge 2:基于神经网络的方法受到更多来自虚假特征的影响,在这里,虚假特征被指定为与事件在语义上类似的潜在信息,但实际上并非如此
Prison authorities have given the nod for Anwar to be taken home later in the afternoon.
Trigger: taken. Event Type: Transport
实验设置与结果
数据集:ACE2005,TAC-KBP2015
实验结果
- 触发器识别
- 事件分类
- embedding类型
- 适应性