URF4CCT: A Text Understanding Framework for Chinese Telecom Fraud Cases

With the increasing prevalence of telecom fraud cases in recent years, there is a growing need for efficient and accurate text understanding techniques to aid police officers in investigating and combating these criminal activities. However, the the complexity of Chinese case texts, the absence of a unified framework for the domain, and the reliance on specialized data annotations pose significant challenges in extracting and representing key information from such texts. In this interdisciplinary study, we propose a unified representation framework for Chinese telecom fraud case texts named URF4CCT using pretrained language models. By employing prompt technology with prior vocabulary enhancement module, URF4CCT improves the model’s ability to capture the unique characteristics of telecom fraud case texts. Experimental results demonstrate the superior performance of our proposed method in low-resource and few-shot scenarios. This research contributes to the development of effective text understanding techniques for telecom fraud cases and facilitates police officers’ investigation efforts in online criminal activities.