English Abstract
Abstract :
As the 4th Industrial Revolution advances and the shift towards a digital workforce intensifies, automation and cybersecurity have become more critical than ever for business enterprises. Robotic Process Automation (RPA) and its AI-driven counterpart, Intelligent Automation (IA), are soft targets for cyberattacks as many of these bots handle clients' sensitive personal data. Many current IA applications suffer from poor logging and auditing practices, as attackers can easily manipulate or delete activity logs to cover their tracks after exploiting the IA bot. This thesis proposes a Blockchain-based Intelligent Automation (BCIA) framework to combat digital fraud and data leakage. By integrating Blockchain with Intelligent Automation, the BCIA Bot Framework automates invoice processing tasks by utilizing Optical Character Recognition (OCR) while securing the bot's activity logs and payment processing receipt evidence onto a BC ledger. The Design Science Research Methodology (DSRM) was applied to guide the development and evaluation of the BCIA framework. This approach ensured a systematic exploration and implementation of the technology. A scenario-based strategy was adopted to simulate real-world applications of the framework for small businesses automating invoice processing in two critical areas of combating data leakage and digital fraud. The BCIA bot successfully stored the activity logs and payment process receipt evidence on the BC at different transaction volumes representing small enterprises (50 and 100 invoices), medium-sized enterprises (500 invoices), and large enterprises (1000 invoices). As for the security framework aspect, despite the BCIA Bot Framework utilizing a public blockchain and a cloud service infrastructure, while the Block-DEF and Blockchain Forensic Logging (BCFL) frameworks utilized private blockchains, the BCIA Bot Framework was able to create a solid foundation for public blockchain IA bot development that provides secure evidence creation and retrieval through a cloud ecosystem. In comparing the system performance metrics of three BC testnets (Sepolia ETH, Solana Devnet, and BNB Chain), which include network throughput, block sizes, transaction fees, transaction latency, and finality time on the BCIA bot, BNB Chain delivered the most efficient performance and cost-effective solutions for small, medium, and large businesses. The BNB-BCIA bot achieved a high average throughput of 47 transactions per minute, an average finality time of 9.3 seconds, and low transaction costs, achieving SUSD 4.20 to process 1000 invoices.
