AI-based Analysis and Classification Enhancement for e-People Civil Complaint Data

Overview

The number of civil complaints filed through e-People (People's Online Petition and Discussion Portal) is continuously rising, with a high volume of simple and repetitive inquiries adding to the administrative burden. Furthermore, the low accuracy of the current system's agency recommendation feature has led to an increased workload for personnel in charge.
To address these issues, the project aims to establish an AI-driven complaint analysis and automation system that enhances complaint summarization, time-series clustering, and agency recommendations by utilizing semantic reasoning and automatic classification technologies. With the Anti-Corruption and Civil Rights Commission (ACRC) as the primary demand organization and VAIV Company as the AI technology provider, this project seeks to improve response efficiency through systematic data analysis and automated classification models.

Project description

This project focuses on developing summarization technology to automatically extract key information from e-People (People's Online Petition and Discussion Portal) data and establishing a monitoring system by applying time-series clustering. Furthermore, it aims to develop a text summarization and agency classification model that provides significantly higher accuracy compared to existing systems. The specific procedures are as follows:

• Redefining Data Analysis & Classification: Improving the complaint classification system to overcome existing limitations and ensure more accurate recommendations for agencies and departments.
• Training Data Construction & Model Development: Building an AI training dataset based on 14,000 complaint records from specific regions and periods to train semantic reasoning and auto-classification models.
• Summarization & Agency Recommendation Model: Utilizing AI-driven summarization models to extract core content and developing a model that automatically recommends the appropriate handling organization.
• Module Integration & Automation Framework: Integrating the summarization and recommendation models into the system to automatically deliver summarized results and recommended agency information to the personnel in charge.

Key technological features

• AI Training Data Construction: Established a comprehensive AI training dataset leveraging 14,000 cases of civil complaint records to ensure model accuracy and reliability
• Semantic Reasoning & Auto-Classification: Developed an advanced complaint analysis and automatic recommendation model by utilizing AI-driven semantic inference and automated classification technologies
• Module Integration & Automation: Built a seamless module integration framework and an automated system architecture to optimize overall operational workflow

Expected outcomes

By introducing an AI-based complaint analysis and automation system, it is expected to increase the efficiency of complaint processing and alleviate the workload of personnel. Enhanced response speed through AI summarization and clustering allows the system to automatically extract key information and group similar complaints, enabling more rapid responses. The goal is to improve the agency recommendation accuracy to over 90%, ensuring that complaints are promptly assigned to the correct organization. This automated system optimized by AI technology will ultimately accelerate digital innovation in administrative services.

Based on the results of this pilot project, there are plans to upgrade the complaint handling systems of e-People and related organizations using AI. This initiative aims to create an environment where citizens can receive faster and more accurate responses to their inquiries.

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=29195