E-learning atau pembelajaran online didefinisikan sebagai aplikasi pembelajaran formal dan informal yang dilakukan hanya dengan menggunakan teknologi, sehingga siswa dan pengajar dapat melakukan proses belajar mengajar melalui media elektronik. Adanya review ulasan pada aplikasi Ruangguru dan Zenius dapat membantu perusahaan memahami opini pelanggan terhadap produk atau layanan. Penelitian ini bertujuan untuk mengetahui opini pelanggan terhadap produk atau layanan dari beberapa aplikasi diantaranya Ruangguru dan Zenius dengan melakukan analisis sentimen yang diambil dari ulasan (review) Website_Google_Play. Pemodelan yang digunakan ialah Long Short-Term Memory (LSTM) dengan ekstraksi fitur Word2vec. Dataset dikumpulkan dengan teknik scraping yang menggunakan bahasa pemrograman Python dan Google scraper. Hasil pengumpulan data aplikasi Ruangguru sebanyak 147.468 dan Zenius sebanyak 22.744 opini terbagi dalam tiga kelas sentimen yaitu positif, negatif dan netral. Model evaluasi menggunakan classification report dengan mengahsilkan empat nilai metrics yaitu akurasi, recall, presisi dan f1-score. Hasil dari metrics dengan data imbalance pada aplikasi Ruangguru memiliki akurasi sebesar 89%, Presisi 88%, Recall 89%, f1-Score 83% sedangkan aplikasi Zenius mendapatkan akurasi sebesar 84%, Presisi 82%, Recall 84%, f1-Score 78% sedangkan pada data balance Ruangguru memiliki akurasi sebesar 46%, Presisi 57%, Recall 46%, F1-Score 42% dan aplikasi Zenius memiliki akurasi sebesar 46%, Presisi 78%, Recall 33%, F1-Score 17%. Penggunaan data imbalance dan data balance memiliki perbedaan, dimana pada data imbalance memiliki akurasi yang tinggi sedangkan pada data imbalance memiliki akurasi rendah dikarenakan penggunaan smote pada representasi vektor word2vec dapat mengakibatkan interpolasi yang tidak tepat dan menyebabkan penyebaran sampel menjadi kurang relevan dalam ruang vektor.


Deskripsi Lain

E-learning online learning is defined a formal and informal learning application that is conducted solely using technology, enabling students and educators to engage in the teaching and learning process through electronic media. The existence of reviews and feedback on the Ruangguru and Zenius applications can help companies understand customer opinions regarding their products services. This research aims to determine customer opinions about the products services of several applications, including Ruangguru and Zenius, by conducting sentiment analysis based on reviews from the Google Play website. The results of this study serve a reference for potential users to choose a platform, allowing them to identify the most influential aspects of the e-learning applications Ruangguru and Zenius. The sentiment analysis in this study involves several preprocessing stages, such case folding, data cleaning, tokenization, stopword removal, stemming, and content cleaning, before moving to the classification stage. The skip-gram model in Word2vec architecture is used to create vector representations of each word (word vector) in the corpus. The sentiment class determination process is carried out using the Long Short-Term Memory (LSTM) method. The research titled "Sentiment Analysis of E-Learning Applications Using Word2vec and LSTM (Long Short-Term Memory)" provides classification results using imbalanced data. Ruangguru application achieved an accuracy of 89%, Precision of 88%, Recall of 89%, and F1-Score of 83%. Meanwhile, the Zenius application obtained an accuracy of 84%, Precision of 82%, Recall of 84%, and F1-Score of 78%. On the balanced data, the Ruangguru application achieved an accuracy of 46%, Precision of 57%, Recall of 46%, and F1-Score of 42%, while the Zenius application had an accuracy of 46%, Precision of 78%, Recall of 33%, and F1-Score of 17%. The usage of imbalanced data and balanced data showed differences. Imbalanced data achieved higher accuracy, whereas balanced data had lower accuracy due to the use of SMOTE (Synthetic Minority Over-sampling Technique) on word2vec vector representation, which resulted in inaccurate interpolation and less relevant sample distribution in the vector space.

• Difuusi Inovasi e-Learning din Perguruan Tinggi
• USING AN OPEN-SOURCE SOFTWARE TO CONSTRUCT AN E-LEARNING SERVICE
• Telaah Teknis terhadap Kinerja Mutu Perkerasan Jalan Nasional dan Propinsi
• USING AN OPEN-SOURCE SOFTWARE TO CONSTRUCT AN E-LEARNING SERVICE
• Telaah Teknis terhadap Kinerja Mutu Perkerasan Jalan Nasional dan Propinsi