基于中文触发词指导的越南语新闻事件检测

Vietnamese News Event Detection Based on the Guida nce of Chinese Triggers

触发词的识别在事件检测任务中起着至关重要的作用。目前没有越南语触发词标记语料,而中文触发词标记语料较为丰富,根据表达相同观点但语言不同的句子通常有相同或相似的语义成分这一特征,该文提出一种基于中文触发词指导的越南语新闻事件检测方法。首先采用对抗学习的方法将两种语言映射到同一语义空间下,然后将映射后的中文触发词嵌入指导模型识别越南语新闻中的触发词信息,最后进行事件类型的分类。通过在越南语新闻事件检测的实验结果表明,在中文触发词指导下的越南语新闻事件检测取得了较好的效果。

The recognition of triggers plays an important role in detecting events.There is no Vietnamese triggers marking corpus at present,but the Chinese triggers marking corpus is relatively rich.Based on the observation that sentences conveying identical idea but in different languages usually have the same or similar semantic components,the paper puts forward a detection method of Vietnamese news event guided by Chinese triggers.Firstly,the two languages are mapped into the same semantic space by adversarial learning.Then the Chinese triggers embeddings are mapped into the guidance.model to identify the triggers information in Vietnamese news.Finally,the events types in Vietnamese news are classified.The experimental result shows that the Vietnamese news event detection a-chieves a better performance on the guidance of Chinese triggers.