AI Audit Service for Detecting Public Sector Misconduct

Overview

New types of employee misconduct are discovered every year in unexpected areas. To prevent such incidents within the organization, SR (Suseo High Speed Railway) aims to derive similar cases based on news articles and historical audit reports. By providing an AI-driven Audit Report Support Service, SR intends to reduce the administrative burden on auditors and support rapid decision-making regarding the analysis of past precedents and the appropriate level of disciplinary action.

Project description

The project utilized internal and external data—including Board of Audit and Inspection (BAI) reports, public institution findings from ALIO, media/SNS content, and internal company regulations—as AI training data for the audit chatbot and report generation models.

The core service features an AI Audit Chatbot that provides comprehensive responses, including summaries of similar audit cases, recommended disciplinary actions, and overall evaluations based on user queries. Additionally, the system supports the AI-powered drafting and editing of audit reports, allowing auditors to generate initial drafts automatically and refine them with AI assistance. This integration ensures that even complex misconduct cases are handled with consistency and based on factual precedents.

Key technological features

• Data Collection & Pre-processing: Removal of whitespaces and noise, standardization of formats, and construction of search indexes
• OCR (Optical Character Recognition): Extraction of text data from images or PDF documents to secure usable training datasets
• Model Tuning: Hyperparameter optimization, including the adjustment of learning rates and batch sizes, to ensure peak model performance

Expected outcomes

Through this service, SR expects to significantly enhance operational efficiency and reduce labor costs associated with manual auditing processes. These improvements will ultimately lead to enhanced management efficiency for the public institution. By leveraging AI to analyze past cases and automate documentation, the organization can respond more proactively to misconduct risks, ensuring transparency and fairness in administrative actions while maintaining high standards of institutional integrity.

Launch year: 2024

Attribution

Case study courtesy of the National Information Society Agency, Republic of Korea, Apolitical's Content Partner. Sourced from Use Cases of Public AI Service, Vol.2, https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do?cbIdx=31975&bcIdx=29182&parentSeq=29230