AI for Finance

• Main Author: Phoe, Chuan Bin
• Other Authors: Erik Cambria
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Document and text processing
• Online Access: https://hdl.handle.net/10356/158241
• Institution: Nanyang Technological University

The rise in prominence of cryptocurrencies have led to increased volatility and trading in cryptocurrency exchanges. Financial institutions are now embracing the use of alternative data especially towards social media commentary to increase their investment returns. The rationale is based on behavioural finance which proved that financial decisions are significantly driven by emotion and mood. As such, sentiment analysis of financial microblogs have been getting increased attention. In this project, I will be leveraging on the use of Text Mining and NLP techniques to better predict the financial sentiment of social media cryptocurrency content. We will take both Symbolic and Sub Symbolic approaches in tackling this problem using lexicons and learningbased language models respectively. Our results show that the proposed final hybrid architecture outperforms individual lexicons in the current literature and state-of-the-art deep learning methods for this sentiment classification problem.