Analyzing Liquidity in Bank with Distributed Knowledge Based Systems A Data Driven Modelling Approach
Abstract
Distributed knowledge-based systems can provide banks with enhanced liquidity analysis based on their data integration and analytical capabilities, so it is necessary to implement these systems for improved liquidity management. Financial professionals are accustomed to using technical financial language to express their strategies and decisions. In order to obtain accurate liquidity assessments from extensive financial reports, it is necessary to apply advanced data processing techniques to ensure precise analysis. Therefore, the study introduced regression analysis and correlation analysis for evaluating liquidity variables. At the same time, the study introduced TF-IDF and regression analysis for processing financial data and introduced a hybrid model combining regression and TF-IDF techniques based on banking data, thereby constructing a distributed knowledge-based system. The innovation of the research lies in the integrated combination of traditional regression and correlation feature weight calculation methods with advanced TF-IDF modeling to enhance liquidity prediction, thereby improving the accuracy and reliability of liquidity management processes. At the same time, combining the advantages of regression analysis and TF-IDF, it effectively processes financial data, extracts relevant liquidity information from it, and achieves more accurate liquidity forecasts. The outcomes indicated that the precision of the classification model combining regression and TF-IDF based on banking data was significantly high, the recall rate was as high as 92%, and the F1-score was 93%. Meanwhile, the baseline model's accuracy was only 85%. In addition, the distributed knowledge-based system, which adopted a data-driven modeling approach, had an average of 1,000 daily visits per person, with an effective browsing time of 5 minutes per person, and a daily browsing page count of 20 per person. This indicates that the data-driven modeling approach has significant practical application effects and provides robust technical support for current research in the field of banking liquidity management.