AI for business intelligence

• Main Author: Goh, Jia Hui
• Other Authors: Erik Cambria
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence
• Online Access: https://hdl.handle.net/10356/154294
• Institution: Nanyang Technological University

A robust stock analysis framework should operate without failure and produce positive results despite changing market conditions and unforeseen circumstances, for stock market analysts to better allocate their assets and reduce possible asset management vulnerabilities. At present, the predictions from stock market analysts are usually based on the opinions expressed in the news. These circumstances give rise to the need to build a Natural Language Processing Model to easily and effectively extract sentiments from speech or in this case, textual expressions. However, the sentiment extracted from the computer may not be the most accurate as modelling linguistics is a challenging task. The accuracy of the sentiment analysis can be further improved with the addition of semantic techniques. Semantic technologies have revolutionized the way systems integrate and share data, enabling computational agents to reason about information and infer new knowledge. This project is about collecting non-quantifiable data such as financial news article headlines about a business and predicting its future stock trend based on the news sentiment classification with the assumption that sentiment from news articles influences the stock market. This is an attempt to study the relationship between the sentiment detected from news titles, stock trends derived from historical stock data and the correlation of stocks. Linear Discriminant Analysis(LDA), Support Vector Machines(SVM) and Gated Recurrent Unit(GRU) are used to depict the relationship of news articles title and stock market movements.