Author(s): Farhana Rahman, Anisur Karim and Nusrat Jahan Akter

Abstract: The rapid expansion of financial services in Bangladesh has created unprecedented demands on large-scale financial databases, where efficient query optimization is critical for ensuring performance, accuracy, and customer satisfaction. Traditional indexing methods, while foundational, often struggle to cope with the dynamic, high-volume, and heterogeneous nature of financial transaction data. This study investigates the effectiveness of adaptive AI-driven indexing models in enhancing query optimization within the Bangladeshi financial sector. Transaction-level datasets from multiple financial institutions were analyzed, incorporating both structured relational data and unstructured log files. Conventional indexing structures such as B-trees and hash indexes were benchmarked against AI-driven models integrating reinforcement learning and neural network frameworks. Statistical analyses, including paired t-tests and ANOVA, revealed significant reductions in query latency, ranging from 30-37%, alongside improvements in throughput, CPU efficiency, and index maintenance times. Although a modest increase in memory usage was observed for transactional workloads, the overall system efficiency gains far outweighed these costs. The results highlight that adaptive AI-driven indexing addresses the limitations of static indexing structures while also providing a scalable, cost-effective, and operationally efficient solution for financial databases in emerging economies. Practical recommendations are proposed for integrating AI-driven indexing into banking systems, including phased implementation, capacity-building initiatives, and hybrid deployment strategies tailored to resource constraints. This study contributes both empirical evidence and practical guidance, positioning adaptive indexing as an essential enabler of digital transformation in the financial sector of Bangladesh.

DOI: 10.33545/27076636.2025.v6.i2a.124

Pages: 151-155 | Views: 84 | Downloads: 41

Download Full Article: Click Here