K-Means Algorithm and Levenshtein Distance Algorithm for Sentiment Analysis of School Zonation System Policy

Abstract

Equity and quality of education must be guaranteed in the national education system. To that end, the government issued a new student admission policy with a zoning system. To ensure the implementation of new student admissions (PPDB), the zoning system needs to be evaluated for community responses. However, evaluation using conventional techniques still has limitations. Sentiment analysis is a new approach to explore computing-based opinion. In this paper, we conduct a sentiment analysis of the new student admissions system (PPDB) zoning policy. We identify two types of sentiment namely positive and negative. We used the Levenshtein Distance algorithm for word normalization and clustered using the K-Means algorithm. The results of clustering are classified based on the confusion matrix. The data sources that we use are taken from 200 comments on Facebook and Youtube channels. The results obtained from public sentiment towards this policy are more negative sentiments than positive sentiments. The results obtained from the accuracy of the K-Means algorithm are 84%, while the combination of the k-means algorithm with Levenshtein distance reaches 90% accuracy.