Sentiment Analysis towards Cryptocurrency and NFT in Bahasa Indonesia for Twitter Large Amount Data Using BERT
Year: 2023
Type: article
Abstract: Sentiment analysis is one part of Natural Language Processing (NLP), where the system can understand the form of text. Sentiment Analysis itself requires an Artificial Intelligence (AI) algorithm. One of them is the Bidirectional Encoder Representation from Transformers (BERT), which can assess positive or negative sentiment, especially when discussing "Cryptocurrency" and "Non-Fungible Token (NFT)" because in recent years. The discussion on these two topics has been widely discussed. On social media such as Twitter. In this study, a BERT model was created to assess sentiment analysis on "Cryptocurrency" and "NFT", by utilizing data crawling and pre-processing using Rapidminer (student version) with 86% accuracy and 87% precision. The results were obtained by selecting suitable hyperparameters such as learning rate 1e-7, epoch 5, and batch size 32. The results also increased from 2% accuracy to 3% precision. Further research needs to be done to improve the BERT model, not just an increase in the pre-processing part. In further research, the authors suggest combining several models or updating the pre-processing.
Institution: Binus University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7217101894
