Project & Service

Project Introduction

This system for generating administrative enforcement documents, referred to as the "Enforcement Model," provides comprehensive guidance throughout the administrative enforcement process and assists with high-quality document drafting, significantly improving the efficiency and quality of enforcement work. Using extensive historical administrative enforcement data, the system trains a large language model specifically for the administrative enforcement domain to enable high-quality, intelligent generation of enforcement documents. This addresses challenges in the enforcement process, such as complex workflows, inconsistencies in enforcement practices, and varying document quality.

Key Research 

· Data Collection and Cleaning: Gather key data from the administrative enforcement field, including enforcement documents, legal provisions, judicial interpretations, case analyses, and other legal business data, to ensure diversity and representativeness.

· Model Training and Optimization: Specialize large language model training for administrative enforcement scenarios, enabling multi-category, multi-scenario legal Q&A, document generation, and intelligent analysis.

· System Development and Application: Develop interfaces and a user interface that provides visual presentations and automated report generation, offering enforcement personnel intelligent assistance tools.

· Risk Identification and Early Warning Module: Create a module to help administrative enforcement scenarios automatically analyze and alert on risk factors. The project plans to use high-performance computing resources, along with distributed computing and cloud storage technologies, to ensure efficient and stable model training and deployment.

Main Outcomes

Application Scenarios and Functions

Administrative enforcement personnel can use photo or voice input to upload critical information from the enforcement process to the model. The model uses image recognition algorithms to accurately identify key details, such as information about the involved parties, enforcement personnel, key enforcement events, evidence, and enforcement actions, allowing for a complete administrative enforcement information flow analysis. By training the model on evidence, events, parties, and other critical data points, it can intelligently generate high-quality documents and legal references. The standardized enforcement workflow simplifies complex processes, ensuring each step meets regulatory standards, achieving full automation of administrative enforcement.

Collaboration Model

Shenzhen Research Institute of Big Data collaborates with various judicial and administrative departments like Justice Bureau of Shenzhen Municipality.