A Systematic Review on Architecture, Applications, and Challenges of Large Model-Driven Intelligent Business Information Systems

Abstract

Against the backdrop of rapid IT development, large, model-driven intelligent business information systems have become a research focus. Integrating large-scale deep learning models, they have significantly boosted business intelligence. However, with explosive data growth, traditional systems face severe challenges in processing efficiency and decision-making quality. Large models exhibit powerful NLP and image recognition capabilities, achieving technological leaps and expanding commercial possibilities. By efficiently adjusting algorithms in resource-constrained environments, they meet diverse business needs, enhancing both data processing efficiency and decision intelligence. Yet, challenges persist. Key issues including data privacy, security, model interpretability, and system compatibility remain urgent to solve. In today's era of increasing focus on commercial secrets and user privacy, ensuring safe, efficient, and rights-respecting data processing and intelligent decision-making with large models has become an important topic in academic research and industrial practice. To address such problems, the academic community has explored various solutions. By building a dynamic data governance system, designing a highly transparent model architecture, and promoting cross-subject collaboration, it has laid a solid theoretical foundation and provided practical guidance for the application of large models in business intelligence. From the perspective of existing research, the intelligent business information system empowered by large models is not merely a technical upgrade, but an in-depth reconstruction of the business operation model. Its transformative impact far exceeds the traditional management paradigm, promoting a fundamental shift in business decision-making logic from manual experience judgment to data-driven scientization.