Development of an AI-Based Solution for Recommending Types of Industrial Accidents and Employment Insurance Programs

Overview

The business type (industry category) of a workplace is critical benchmark information for setting premium rates for workers' compensation insurance. Due to recent changes in the industrial structure, determining business types has become increasingly complex. This necessitates overcoming the limitations of relying solely on the experience of existing staff members for business type classification.

To address this, a project was launched in 2023 to develop and validate a business type recommendation solution. This initiative supports an AI expert consortium to enhance operational efficiency by recommending the optimal business type through AI implementation.

Project description

The Ministry of Employment and Labor and the Korea Workers' Compensation & Welfare Service are developing and demonstrating an AI solution equipped with a function to recommend types of industrial accident and employment insurance programs. This is achieved by building and training AI learning data using information from various documents and other data held by these organizations.
*(Structured) Total of 30 types of business category determination data / (Unstructured) Various submitted documents [in file formats (HWP, PDF, image)] that provide workplace basic information and confirm workplace conditions

The AI-based Work Injury and Employment Insurance Business Type Recommendation Solution utilizes various reference materials and data necessary for determining business types to recommend appropriate Work Injury and Employment Insurance categories. It provides a service that explains the rationale behind each recommendation in detail to enhance user understanding of the results.

Processed data volume

Key technological features

• AI OCR Model: Building AI training data through processing and refining years of accumulated workers' compensation and employment insurance application data (business registration certificates, financial statement certificates, etc.)
• AI Recommendation Model: Implementation of features such as recommending types of workers' compensation and employment insurance businesses using AI technologies (e.g., BERT).

Expected outcomes

The AI-based industrial accident and employment insurance business type recommendation solution achieved 92% accuracy in industry recommendations through development and performance enhancement starting in 2023. Furthermore, the AI solution underwent a pilot application for Korea Workers' Compensation & Welfare Service investigators by the end of October 2024. When reviewing multiple documents to change or determine workers' compensation and employment insurance program types, it reduced the time per case from a maximum of two months to just 30 seconds. The introduction of this AI solution is expected to enhance public trust by securing the fund's financial soundness and improving administrative efficiency, while also ensuring the stable operation of workers' compensation and employment insurance programs.

Launch year: 2023

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.1,  https://eng.nia.or.kr/site/nia_eng/ex/bbs/View.do?cbIdx=31975&bcIdx=28997&parentSeq=29015