DETECTING WEAK SIGNALS THROUGH TEXT MINING FOR COMPETITIVE INTELLIGENCE

This thesis presents a novel approach to detecting weak signals in financial data for competitive intelligence using an automated system designed through the lens of a design science research methodology. Composed of three integrated modules - Text Mining, Weak Signal Detection, and Competitive Intelligence Integration - this system transforms raw, unstructured data into structured text, extracts weak signals as potential early indicators of emerging trends within the financial sector, and aligns these signals with a comprehensive competitive intelligence framework. Aided by techniques such as dynamic topic modelling and sentiment analysis, the system adeptly navigates the complexities of financial data, identifying significant trends and anomalies. Despite computational limitations and the challenges of evaluating robustness and noise in the data, the system has proven efficient in processing high-velocity financial data, adaptable to diverse data types, and capable of real-time analysis. Including a feedback loop allows the system to evolve and continually improve, enhancing its accuracy over time. Moreover, the study explores future research directions, including incorporating financial statements, improving user interaction and decision-support tools, expanding the analysis to other sectors, and developing strategic financial insights extraction and competitive intelligence report generation capabilities. This research demonstrates the potential of an automated weak signal detection system in revolutionising financial data analysis and strategic decision-making in the rapidly evolving financial sector. This study contributes to the ongoing discourse on the fusion of text-mining techniques and competitive intelligence in the finance sector, offering a roadmap for future research and technological development