Abstract:

With the continuous advancement of Artificial Intelligence (AI)-based recognition technologies and the deepening digital transformation of government services, AI recognition in public-sector applications is expected to experience substantial growth. Contract data entry and invoice recognition constitute key components of the contract execution and payment phases within the government procurement management workflow, serving as crucial steps to ensure compliance and operational efficiency. This paper proposes an intelligent form automation tool tailored for government service scenarios. The tool integrates optical character recognition, semantic fuzzy matching, and vision-driven robotic process automation technologies, achieving full-process automation—including field extraction, semantic error correction, and cross-system data entry—through multimodal collaboration. Experimental results demonstrate that the tool attains an average overall accuracy of 99.2% across tasks such as contract field parsing, invoice information extraction, and result entry, significantly reducing manual data-entry costs and improving the efficiency of administrative workflows. The findings provide a scalable and practical technical pathway for deploying lightweight AI solutions in grassroots government contexts.

Key words: artificial intelligence, CV, RPA, OCR, intelligent recognition