Abstract:
This research addresses the integration of Robotic Process Automation (RPA) and
Machine Learning (ML) to enhance the efficiency and accuracy of document
processing in various business environments. The study identifies the challenges
associated with traditional document processing methods, which are often time consuming and error-prone, and explores how the synergistic application of RPA and
ML can mitigate these issues, thereby revolutionizing document handling practices.
The methodology adopted involves a combination of RPA to automate routine
document handling tasks and ML to intelligently classify and extract data from a wide
range of document formats. This integration was prototyped using an advanced ML
model trained on a labeled dataset to recognize and categorize document contents
accurately to finance document. The RPA component was developed to seamlessly
fetch, process, and route documents based on the insights derived from the ML model,
ensuring a streamlined workflow.
Initial evaluation of the prototype demonstrated an 95% classification accuracy for
financial documents, supported by an AUC-ROC score of 0.91. The confusion matrix
analysis revealed a 3% false-positive rate. Additionally, the system achieved a real time processing capability of up to 80 pages of document per minute, highlighting the
robustness and scalability of the integrated RPA and ML solution.
