The pandemic caused by the 2019 coronavirus has revitalized telemedicine as information and communication technology-based health services and as a medium for doctors' services in diagnosing, treating, preventing and evaluating health conditions. One of the telemedicine service applications in Indonesia is Alodokter, Halodoc, KlikDokter, SehatQ and YesDok. Previous research on the same domain, namely applications telemedicine uses machine learning to perform sentiment modeling. This research performs sentiment analysis using the BiLSTM method (Bidirectional Long Short-Term Memory) which can better represent contextual information and can read user feedback information in both directions. Then sentiment analysis is described explicitly to identify topics from user sentiment using LDA (Latent Dirichlet Allocation). User feedback was collected on August 14, 2022 which was obtained in the five applications totaling 244,098. The results of the analysis on feedback obtained were 112,013 positive sentiments, 34,853 neutral sentiments and 97,228 negative sentiments. The BiLSTM and Word2Vec models used have a good performance in classifying sentiments, namely 95%, while the topic modeling for each sentiment has a coherence value of 0.6437 on positive topics, 0.6296 neutral sentiments and 0.6132 negative sentiments.
Copyrights © 2023