An Improved Tri-Training Based Named Entity Identification Approach for Legal Knowledgebase of Properties Involved in Criminal Cases

The legal knowledgebase of properties involved in criminal cases aims to automatically complete the knowledge fusion related to the disposal of case-involved properties in criminal cases based on the existing laws and regulations, and provide support to the case handlers of law enforcement and judicial units such as public prosecutors and law enforcement in judicial practice. In this paper, we explore semi-supervised learning based on tri-training for named entity identification of case-related property knowledgebase based on a small amount of annotated data to reduce the workload of corpus annotation. To overcome the problem that the original tri-training method tends to degenerate into a single learner when the amount of training data is small, we propose an improved tri-training method. Experiments show that our proposed method can effectively improve the performance of named entity recognition compared to the traditional tri-training scheme.