Automating data entry for an online biomedical database: a document image analysis application

Creating online bibliographic databases from paper-based journal articles continues to be heavily manual. An R&D center at the United States National Library of Medicine (NLM), is developing systems for automating the extraction of information from biomedical journals to create bibliographic records in MEDLINE(R), NLM´s premier online database used worldwide. The first phase of this project has resulted in a system that involves scanning and converting (by OCR) the abstracts that appear in journal articles, while keyboarding the remaining fields. A second generation system is being designed to scan/OCR other fields such as author names, institutional affiliations, page numbers, article titles and others. This system will employ scanning and OCR as well as modules that automatically zone the scanned pages, identify the zones as particular fields, and reformat the field syntax to adhere to conventional practice in MEDLINE