Financial sentiment analysis in BIST100 companies’ annual reports: A comparision of dictionary based and deep learning based methods

Date

2025-1

Author

Tür, Ahmet Şamil

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

150
views

162
downloads

Cite This

Within the complicated finance environment, understanding market sentiments is crucial for investors, financial analysts, and decision-makers. This study looks at sentiment analysis, a powerful approach supported by natural language processing and machine learning. Carefully gathered, a comprehensive dataset of BIST100 companies' annual reports creates the foundation of this research, allowing detailed analyses that extract insights from the combination of emotions and financial data. Using different sentiment analysis methodologies, from dictionary-based to deep learning-based, this research offers a broad view of sentiment patterns embedded within the text of financial reports. The calculated sentiment scores are compared and contrasted, revealing the strengths and limitations of each approach. This comparative analysis shows the capacity of each method to capture nuances within the complex financial language. Moreover, these sentiment scores show a fascinating connection with Türkiye's GDP growth rates over time, forming a clear link between economic patterns and expressed sentiments. The observed correlation between financial performance and sentiment trends offers a view into how market dynamics influence the emotional themes in annual reports. By improving our understanding of sentiments in financial documents, this study not only supports more informed decision-making but also builds a base for future research. As sentiment analysis continues to develop, this work opens the way for integrating emerging methodologies, showing the potential of sentiment analysis in shaping the future of financial analysis and decision-making.

Subject Keywords

financial sentiment analysis, natural language processing, deep learning, annual activity reports, BIST100

URI

https://hdl.handle.net/11511/113400

Collections

Graduate School of Social Sciences, Thesis