<![CDATA[This research aims to analyze public sentiment regarding services at the Department of Population and Civil Registration as an objective evaluation instrument for service quality through a series of methodological stages, beginning with text preprocessing including cleansing, folding, tokenizing, filtering, and stemming to reduce noise in the scraped textual data. Feature extraction was performed using Term Frequency-Inverse Document Frequency (TF-IDF) to determine the significance of each term within the documents, while the classification process was implemented using the K-Nearest Neighbor (KNN) algorithm by experimenting with various nearest neighbor (k) values to identify the most optimal model parameters. Based on a comprehensive evaluation using a confusion matrix, the model achieved peak performance at k=3 with an accuracy rate of 92.8%, although significant limitations were identified in predicting negative sentiments, only 6 of the 14 data points were correctly classified. This indicates a classification bias triggered by data imbalance, a common challenge in text mining is that the number of positive reviews far outnumbers the number of negative reviews. Qualitatively, although reviews were dominated by public appreciation, critical complaints were still found regarding perceived slow service duration and overlapping bureaucratic complexities. These findings emphasize that while the model possesses high quantitative accuracy, a thorough evaluation of service quality must focus on strategic transformations that improve system efficiency and information transparency to bridge the gap between community expectations and the reality of public service delivery.]]>
