Automated Text Recognition and Review System for Enhanced Bidding Document Analysis

This paper aims to extract, analyze, and review various types of material information, such as business licenses and qualification certificates, from electronic documents in bidding processes. These documents may be in various formats such as images, scanned copies, and PDFs. Utilizing CRNN and building upon well-trained models, our method demonstrates strong error correction capabilities, thereby enhancing the text recognition accuracy of the Guizhou Power Grid Company’s bidding documents. The system is designed to automatically extract and scrutinize key information from multi-format bidding documents obtained from diverse sources. It enables bid evaluation experts to accurately identify and locate issues in the essential information presented in various materials within the bidding documents, through a user-friendly visual interface. Experimental results indicate that our system can precisely extract pivotal information from bidding documents across various formats, consequently reducing the review and approval time. This research holds significant value and presents promising application prospects by potentially lowering labor costs, as well as improving the efficiency and fairness of the bidding process.