| dc.description.abstract | Travel using airlines is one of the modes of transportation that is in great demand by the people of Indonesia. The number of airlines in Indonesia is in the spotlight for the public, especially the airline Garuda Indonesia. Various public opinions on the performance of the Garuda Indonesia airline were conveyed through the social media Twitter. However, this matter has not been described and explained in a simple way about the aspects discussed and how big the percentage of sentiment is. Thus, an aspect-based sentiment analysis is needed on public opinion regarding the airline Garuda Indonesia so that it can be identified which aspects are widely discussed and whether these opinions include positive, negative or neutral sentiments. In this study, analysis was carried out using Latent Dirichlet Allocation (LDA) and Multinomial Naïve Bayes (MNB) algorithms. In this study, 1750 tweet data were used. The data is processed for clustering using LDA where 3 aspects are obtained, namely: facilities, services and schedules. Then the data is divided into 1400 training data and 350 testing data with a percentage of 80:20. The data is preprocessed, including: cleaning, case folding, normalization, tokenizing, filtering and stemming. MNB is used to determine the sentiment group. After that, an evaluation of the system was carried out using the confusion matrix and an accuracy value of 80% was obtained. | en_US |
