Joint entity and event extraction model based on multi task deep learning

Abstract

【 Purpose 】  Use the correlation between named entity recognition and event detection tasks , At the same time, it improves the performance of named entity recognition and event detection models .【 Method 】  A joint entity and event extraction model based on multi task learning is proposed MDL-J3E, The model is divided into sharing layers 、 Private layer and decoding layer . among , The shared layer generates common features ; The private layer consists of named entity recognition part and event detection part , Based on the general features, the private features of the two subtasks are extracted respectively ; The decoding layer decodes and outputs the characteristics of the subtask into a tag sequence that conforms to the constraint rules .【 result 】<