Pahlevi, Farid (2024) Pemodelan Topik Dan Analisis Sentimen Berbasis Aspek Terhadap Penggunaan Ganja Pada Video Dokumenter “atas Nama Daun” Di Youtube. Other thesis, Universitas Islam Riau.

Text Farid Pahlevi.pdf - Submitted Version Restricted to Registered users only Download (7MB) | Request a copy

Abstract

Cannabis is a class one narcotic that is prohibited from use for health purposes as described in Law No. 35 of 2009 concerning Narcotics. Cannabis has become a topic of debate and research, even some countries have legalized it for various purposes, especially medical. One of the events of medical cannabis use that was widely discussed in Indonesia was Fidelis Arie. He decided to grow and extract cannabis for his wife's treatment. His wife's condition slowly improved, but was halted by Fidelis' arrest. This incident became one of the segments in the documentary "Atas Nama Daun". This documentary discusses the use of cannabis from various sources' perspectives and has been watched 2.9 million times on YouTube. This research focuses on topic modeling and aspect-based sentiment analysis of comments on the documentary video "Atas Nama Daun" on YouTube. Topic modeling is done with the aim of obtaining topics or aspects on comments using Latent Dirichlet Allocation which is evaluated by coherence score. Next is to perform aspect-based sentiment analysis based on the aspects identified in topic modeling using 4 traditional machine learning approaches, namely Naïve Bayes, K-Nearest Neighbor, Decision Tree, and Logistic Regression which are evaluated by comparing accuracy values. The aspects identified in topic modeling are Legal and Health aspects, specifically topics about the laws and regulations for the use of medical cannabis in Indonesia. The results of aspect-based sentiment analysis show that Logistic Regression is the best model for aspect classification with an accuracy of 84.5% and sentiment classification for Legal aspects with an accuracy of 79%. While Naïve Bayes provides the best performance for sentiment classification for the Health aspect with an accuracy of 77.1%. Overall, the majority of comments criticize or disagree with the law, legislation, or government regarding the use of cannabis and consider the use of cannabis does not cause significant adverse health effects.

Actions (login required)

View Item