ASPECT-BASED SENTIMENT ANALYSIS MENGENAI PEMISAHAN MODE MAGIC CHESS MENJADI APLIKASI MAGIC CHESS GO GO (MCGG) DENGAN METODE LSTM DAN BILSTM

FARHAN NOOR HIDAYATULLAH, . (2026) ASPECT-BASED SENTIMENT ANALYSIS MENGENAI PEMISAHAN MODE MAGIC CHESS MENJADI APLIKASI MAGIC CHESS GO GO (MCGG) DENGAN METODE LSTM DAN BILSTM. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (652kB)
	Text BAB 1.pdf Download (317kB)
	Text BAB 2.pdf Restricted to Registered users only Download (625kB) | Request a copy
	Text BAB 3.pdf Restricted to Registered users only Download (283kB) | Request a copy
	Text BAB 4.pdf Restricted to Registered users only Download (377kB) | Request a copy
	Text BAB 5.pdf Restricted to Registered users only Download (301kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (305kB)
	Text LAMPIRAN.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Farhan Noor Hidayatullah, Aspect-Based Sentiment Analysis Mengenai Pemisahan Mode Magic Chess Menjadi Aplikasi Magic Chess Go Go (MCGG) dengan Metode LSTM dan BiLSTM. Dosen Pembimbing: Dr. Widodo, S.Kom., M.Kom., Murien Nugraheni, S.T., M.Cs. Program Studi Pendidikan Teknik Informatika dan Komputer. Fakultas Teknik. Universitas Negeri Jakarta. 2026. Penelitian ini bertujuan untuk melakukan Aspect-Based Sentiment Analysis (ABSA) pada ulasan pengguna terhadap pemisahan mode Magic Chess menjadi aplikasi mandiri Magic Chess Go Go (MCGG). Evaluasi sentimen difokuskan pada pemodelan berbasis Deep Learning menggunakan arsitektur Long Short-Term Memory (LSTM) dan Bidirectional Long Short-Term Memory (BiLSTM) untuk mengklasifikasikan sentimen pengguna ke dalam enam kelas kombinasi aspek dan sentimen (positif, netral, dan negatif) pada dua aspek utama, yaitu Gameplay/User Experience dan Kebijakan. Penelitian ini menggunakan teknik ekstraksi fitur Word2Vec secara static embedding dan metode Synthetic Minority Over-sampling Technique (SMOTE) secara global untuk menangani ketidakseimbangan kelas pada dataset ulasan. Hasil pengujian menunjukkan bahwa arsitektur BiLSTM secara umum sedikit lebih unggul dibandingkan LSTM dengan perolehan tingkat akurasi sebesar 41%, melampaui performa LSTM yang mencapai akurasi 40%. Pada tingkat aspek, BiLSTM menunjukkan keunggulan pada aspek Gameplay(weighted F1-Score 0,59 berbanding 0,55), namun justru tertinggal dari LSTM pada aspek Kebijakan (weighted F1-Score 0,38 berbanding 0,41). Selain itu, keunggulan performa BiLSTM secara keseluruhan diikuti dengan waktu pelatihan komputasi yang lebih lama (69,74 menit) dibandingkan LSTM (33,85 menit). Hasil analisis sentimen aspek mengungkapkan bahwa sentimen pengguna didominasi oleh respons negatif, terutama pada aspek Gameplayyang mengumpulkan 7.401 ulasan negatif dan aspek Kebijakan dengan 4.628 ulasan negatif terkait pemisahan aplikasi tersebut. Kata Kunci: Aspect-Based Sentiment Analysis, Bidirectional Long Short-Term Memory, Long Short-Term Memory, Magic Chess Go Go, Analisis Sentimen, SMOTE, Word2Vec. ***** Farhan Noor Hidayatullah, Aspect-Based Sentiment Analysis on the Separation of Magic Chess Mode into the Magic Chess Go Go (MCGG) Application using LSTM and BiLSTM Methods. Advisors: Dr. Widodo, S.Kom., M.Kom., Murien Nugraheni, S.T., M.Cs. Study Program of Informatics and Computer Engineering Education. Faculty of Engineering. Universitas Negeri Jakarta. 2026. This study aims to conduct an Aspect-Based Sentiment Analysis (ABSA) on user reviews regarding the separation of the Magic Chess mode into a standalone application, Magic Chess Go Go (MCGG). The sentiment evaluation focuses on Deep Learning-based modeling using Long Short-Term Memory (LSTM) and Bidirectional Long Short-Term Memory (BiLSTM) architectures to classify user sentiments into six combined aspect-sentiment classes (positive, neutral, and negative) across two main aspects, namely Gameplay/User Experience and Policy. This research utilizes the Word2Vec feature extraction technique as a static embedding and the Synthetic Minority Over-sampling Technique (SMOTE) applied globally to address class imbalance in the review dataset. The test results show that the BiLSTM architecture generally performs slightly better than LSTM, achieving an accuracy of 41%, surpassing LSTM's accuracy of 40%. At the aspect level, BiLSTM demonstrates superior performance on the Gameplayaspect (weighted F1-Score of 0.59 versus 0.55), but is outperformed by LSTM on the Policy aspect (weighted F1-Score of 0.38 versus 0.41). Furthermore, the overall superior performance of BiLSTM is accompanied by a longer computational training time (69.74 minutes) compared to LSTM (33.85 minutes). The aspect-based sentiment analysis results reveal that user sentiment is heavily dominated by negative responses, particularly in the Gameplayaspect, which garnered 7,401 negative reviews, and the Policy aspect, with 4,628 negative reviews regarding the application's separation. Keywords: Aspect-Based Sentiment Analysis, Bidirectional Long Short-Term Memory, Long Short-Term Memory, Magic Chess Go Go, Sentiment Analysis, SMOTE, Word2Vec.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Dr. Widodo, M.Kom. 2). Murien Nugraheni, S.T., M.Cs.
Subjects:	Teknologi dan Ilmu Terapan > Teknik Komputer
Divisions:	FT > S1 Pendidikan Teknik Informatika Komputer
Depositing User:	Users 35252 not found.
Date Deposited:	11 Aug 2026 03:57
Last Modified:	11 Aug 2026 03:57
URI:	http://repository.unj.ac.id/id/eprint/69577

Actions (login required)

View Item