A study in applying optical character recognition technology for the Foreign Broadcast Information Service field bureaus

The Foreign Broadcast Information Service (FBIS) collects and disseminates world-wide open-source information for the U.S. government through a collection of 17 field sites, or bureaus, located in cities around the world. Several bureaus collect a large amount of English-language material that is manually rekeyed into a computer database. Since this is a labor-intensive process that prevents some bureaus from meeting their processing requirements, FBIS is interested in applying a system that would make this process more efficient.

From analyzing the requirements for an improved text-entry system and evaluating several alternative solutions in terms of cost and feasibility, a design approach using Commercial Off-The-Shelf (COTS) Optical Character Recognition (OCR) technology is recommended. The technical requirements for using OCR systems at FBIS field sites is presented along with evaluation techniques for choosing cost-effective COTS OCR products. Finally, the requirements for testing alternative OCR system designs under field operating conditions is included to determine the specific range of printed materials effectively processed by OCR.