Harnessing AI and Human Expertise to Create a Reliable Digital Archive

Challenge

The “Dutch Cards” project was launched to digitize 1.7 million handwritten and typewritten Dutch civil records from 85 microfilm reels for archival use. Each card contains critical data, such as family names and registration numbers, with strict accuracy requirements: 98% for all fields and 99.8% for specific fields. The challenge included handling varied formats and mixed handwriting and typewritten text, making high-precision data extraction essential.

DDD’s Solution

To meet the Dutch government’s accuracy standards, DDD used a multi-step process. High-resolution scanning of 85 reels produced clear images, which were enhanced through cropping, de-skewing, and contrast adjustments for reliable extraction. DDD’s specialized software, trained to recognize varied card layouts using AI and rule-based algorithms, accurately mapped fields. OCR technology tailored for Dutch handwriting and typewritten text enabled precise extraction, while trained reviewers performed quality checks and corrected errors to achieve 99.8% accuracy in critical fields.

Transforming Historic Records with High-Tech Accuracy and Human Quality Assurance

Impact

While ongoing, the project now consistently delivers batches meeting the Dutch government’s high standards, ensuring an accurate digital archive of Dutch civil records. This enables the preservation of critical historical data with the precision required for government use.