Komparasi model sentimen analisis berbasis bert pada studi kasus cuitan bertagar #israelpalestinewar = Comparison of sentiment analysis models based on #israelpalestinewar tweets case study

Putra, Angga Gamma (2024) Komparasi model sentimen analisis berbasis bert pada studi kasus cuitan bertagar #israelpalestinewar = Comparison of sentiment analysis models based on #israelpalestinewar tweets case study. Bachelor thesis, Universitas Pelita Harapan.

	Text (Title) Title.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (430kB)
	Text (Abstract) Abstract.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (601kB)
	Text (ToC) ToC.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (606kB)
	Text (Chapter1) Chapter1.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (738kB)
	Text (Chapter2) Chapter2.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter3) Chapter3.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (3MB)
	Text (Chapter4) Chapter4.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)
	Text (Chapter5) Chapter5.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (595kB)
	Text (Bibliography) Bibliography.pdf Restricted to Registered users only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (607kB)
	Text (Appendices) Appendices.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial Share Alike. Download (1MB)

Abstract

X merupakan salah satu media sosial yang mampu menampilkan komentar dengan sedikit atau tanpa filter sama sekali. Ini berarti bahwa tweet atau cuitan akan dapat menampilkan emosi seseorang dalam bentuk yang lengkap. Pada topik penelitian ini, perang antara Israel dan Palestina saat ini menjadi topik hangat di X. Orang-orang telah membicarakan perang ini dan pendapat telah terbagi menjadi pro dan kontra dari masing-masing atau kedua negara. Pendapat ini sangat berharga karena mewakili konflik dan bentrokan pendapat dengan cara yang sangat lugas. Oleh karena itu, hal ini akan membantu dalam pengembangan model analisis sentimen yang dapat melihat bentuk sentimen-sentimen dari dua kubu pendapat yang berbeda. Pengembangan model analisis sentimen menggunakan teknik pemrosesan bahasa alami (NLP) menggunakan transformer sebagai model pra-pelatihan dasar. Model yang akan digunakan adalah Robustly Optimized Bidirectional Encoder Representations from Transformers Approach (RoBERTa) dan Distilled Bidirectional Encoder Representations from Transformers (DistilBert). Penelitian ini akan menggunakan kumpulan data yang telah dikumpulkan dari hashtag #IsraelPalestineWar dari X dengan rentang waktu antara Oktober hingga November 2023. Kinerja model kemudian akan dievaluasi menggunakan metrik evaluasi seperti akurasi, recall, presisi, F1-score dan confusion matrix. Model yang dibuat menghasilkan hasil yang memuaskan dalam evaluasi metrik dengan rata-rata 95% untuk setiap model. Model telah diuji dengan input sampel yang berbeda dari dataset asli dan memberikan output sentimen yang benar. Namun, model memiliki beberapa tantangan dalam memproses emoji, ambiguitas, dan sarkasme. / X is one of many social media platform that allows for comments to be displayed with little to no filtering. This means that tweets can fully express a person's emotions. In the context of this research topic, the ongoing conflict between Israel and Palestine has become a hot topic on X. People have been discussing this war, and opinions have been divided into pro and contra for either or both countries. These opinions are valuable as they represent the conflict and clash of views in a very straightforward manner. Therefore, this will assist in developing a sentiment analysis model that can detect the forms of sentiment from the two opposing sides. The sentiment analysis model development will use natural language processing (NLP) techniques with transformers as the base pre-trained model. The models used are the Robustly Optimized Bidirectional Encoder Representations from Transformers Approach (RoBERTa) and the Distilled Bidirectional Encoder Representations from Transformers (DistilBERT). This research will utilize a dataset collected from the hashtag #IsraelPalestineWar on X, spanning from October to November 2023. The performance of the models will then be evaluated using evaluation metrics such as accuracy, recall, precision, F1-score, and confusion matrix. The models that were created yield satisfactory results in the metric evaluation with an average of 95% for each models. Models have been tested with sample inputs, different from the original dataset and gave correct output sentiments. However, the models have some challenges in processing emojis, ambiguity and sarcasm.

Actions (login required)

View Item