LLMs for Financial Document Processing
DOI:
https://doi.org/10.63282/3050-9262.IJAIDSML-V5I1P116Keywords:
Large Language Models (LLMs), Financial Document Processing, Natural Language Processing (NLP), Multimodal Learning, Financial Text Analytics, Time-Series Forecasting, Risk Assessment, Regulatory Compliance, Decision Support Systems, AI in FinanceAbstract
The conventional methods of financial document analysis have been based on official numerical pointers, without considering the huge narrative data in company reports, regulatory submissions, morale reports, or sector statements. According to the recent progress in the sphere of -informed large language models (LLMs), machines have become capable of processing and encoding financial documents with better accuracy than at any previous point in history. This paper presents a multimodal LLM architecture, which makes use of financial reports to synthesize a narrative and lock it in with a time-series predictor in external downstream trainers that do not use the narrative, namely forecasting and control. As the energy and insurance sectors view of case studies evidence, document processing through LLM does increase the accuracy of prediction in addition, that it makes reading or comprehension superior, contributing to risk identification, harmonization of the rules, and decision-making. The findings highlight the value of LLMs as a methodological roadmap for financial document processing beyond the banking sector.
References
[1] Ang, G., & Lim, E. P. (2022, May). Guided attention multimodal multitask financial forecasting with inter-company relationships and global and local news. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) (pp. 6313-6326).
[2] Bellotti, A., Brigo, D., Gambetti, P., & Vrins, F. (2021). Forecasting recovery rates on non-performing loans with machine learning. International Journal of Forecasting, 37(1), 428-444.
[3] Chan, Y. Y., & Qu, H. (2016, January). Finavistory: Using narrative visualization to explain social and economic relationships in financial news. In 2016 International Conference on Big Data and Smart Computing (BigComp) (pp. 32-39). IEEE.
[4] Emami, H., Dang, X. H., Shah, Y., & Zerfos, P. (2023). Modality-aware Transformer for Financial Time series Forecasting. arXiv preprint arXiv:2310.01232.
[5] Guo, W., Wang, J., & Wang, S. (2019). Deep multimodal representation learning: A survey. Ieee Access, 7, 63373-63394.
[6] Lee, S. I., & Yoo,S. J. (2020). Multimodal deep learning for finance: integrating and forecasting international stock markets. The Journal of Supercomputing, 76(10), 8294-8312.
[7] Liu, Y., Li, Y., & DeGeronimo, B. A. D. (2020). FinBERT: A Pretrained Language Model for Financial Communications. Journal of Computational Finance, 24(4), 15.
[8] Loughran, T., & McDonald, B. (2020). Textual analysis in finance. Annual Review of Financial Economics, 12(1), 357-375.
[9] Lv, X., Xiong, X., & Geng, B. (2023). Increasing the prediction performance of temporal convolution network using multimodal combination input: Evidence from the study on exchange rates. Frontiers in Physics, 10, 1008445.
[10] Rodrigues, F., Markou, I., & Pereira, F. C. (2019). Combining time-series and textual data for taxi demand prediction in event areas: A deep learning approach. Information Fusion, 49, 120-129.
[11] Sardelich, M., & Manandhar, S. (2018). Multimodal deep learning for short-term stock volatility prediction. arXiv preprint arXiv:1812.10479.
[12] Sawhney, R., Mathur, P., Mangal, A., Khanna, P., Shah, R. R., & Zimmermann, R. (2020, October). Multimodal multi-task financial risk forecasting. In Proceedings of the 28th ACM international conference on multimedia (pp. 456-465).
