Implementation of K-Means Clustering in a Web-Based Public Complaint System to Improve Efficiency and Transparency

Abstract

The community complaint system in Sumengko Village still relies on conventional methods such as direct visits to the village office and deliberation forums, making it difficult for village officials to systematically categorize complaints and determine the appropriate handling priorities. This research aims to design and develop a web-based community complaint system integrated with the K-Means Clustering algorithm to automatically group complaints based on the similarity of text characteristics. The system is built using the Laravel framework based on PHP with a MySQL database, and applies text preprocessing stages including case folding, tokenizing, stopword removal, and stemming, followed by TF-IDF weighting to convert the text into a numerical representation before the clustering process is performed. The dataset used consists of 15 complaint documents representing three dominant topics: infrastructure, environment, and health. Validation of the number of clusters using the Elbow Method shows k=3 as the optimal point with the lowest significant SSE value. The clustering results show that the K-Means algorithm converges at the 2nd iteration with the following cluster distribution: C₁ Infrastructure (60%), C₂ Environment (33.3%), and C₃ Health (6.7%), indicating that infrastructure-related complaints are the dominant issue reported by residents. Usability evaluation using the System Usability Scale (SUS) yielded an overall mean score of 79.25, categorized as "Good" with an acceptability level of "Acceptable," confirming that the system is well-received by both village administrators and community members. This system not only serves as a medium for submitting complaints digitally but also as an automatic analysis tool that helps village officials prioritize complaint handling more efficiently and transparently.