Penerapan Latent Dirichlet Allocation untuk Aspect-Oriented Sentiment Analysis Berdasarkan Penilaian Penggunaan Layanan di Perguruan Tinggi

Abstract

Improving higher education academic services according to student needs can be done by analyzing the aspects discussed in online reviews from students and their sentiments. With an increasing number of online reviews, the process of manually analyzing online reviews will require a lot of time and effort, so an automated or semi-automated analysis system is needed to increase efficiency and effectiveness. This study aims to do the task of aspect-oriented sentiment analysis where aspects discussed by students in the review text will be identified using the Latent Dirichlet Allocation method and the review sentiment will be calculated using the sentiment lexicon whether the review is positive, negative or neutral. The aspects of higher education academic services that were evaluated were tangibles, reliability, responsiveness, assurance, and empathy. As a result, the LDA method succeeded in identifying 4 of the 5 aspects evaluated except for the assurance aspect. Aspects of tangibles and reliability are known to be the most discussed aspects by students with sentiments that tend to be positive towards aspect of reliability and sentiment that tend to be negative towards aspect of tangibles.