A Case For Combining OCR And Document Management

High-volume document management and optical character recognition (OCR) are as inseparable as the insurance industry and American International Group Inc (AIG). AIG is a consortium of companies in the financial services and insurance businesses. The four came together in mid-1997. An AIG member company, National Union Fire Insurance Company of Pittsburgh, PA, installed a nationwide document management system to reduce legal research costs.

National Union is a provider of directors' and officers' professional liability and other liability insurance. To provide the best litigation defense for its policyholders, National Union formed a team of attorneys from top law firms - the National Union Panel Counsel.

The Driving Reason For Document Management And OCR
Legal research is time consuming, laborious and expensive. After analyzing cases, National Union found that Panel Counsel firms, working on similar cases, were duplicating research efforts, interviewing similar types of witnesses and searching for identical case precedents. Sharing this research would save time and costs.

Addressing this issue, National Union used Icon Software's IntroSpect to create Brief Base®, a free, online database of legal research. Brief Base, exclusively for use by Panel Counsel firms, is a remotely accessed, stand-alone document server which incorporates various IntroSpect functionalities. One example is the addition of OCR on the front end to translate scanned documents into searchable ASCII text. Brief Base is a scalable application which can use either original e-mail or hard-copy documents to be reviewed by users. It provides National Union with very flexible indexing through unlimited user-defined index fields. Brief Base allows Panel Counsel attorneys to browse images of the latest documents, submitted by Panel Counsel member firms. These documents relate to federal and state securities decisions, opinions, motions, briefs and expert witness testimony.

Says Roy Vincent, assistant vice president, development, Information Systems Group, AIG, "We wanted to develop a sharable library of information that any of our Panel Counsel member firms could use at any time."

How The Solution Is Achieved
Expert testimony reports, case judgments and other files are sent in hard copy or electronic image format to a processing department in New York. There, a librarian prepares them for conversion into a searchable database. All submissions are catalogued into categories: state, federal, expert testimony and appellate. Cover sheets, which include category fields, are available to slot documents into these specific categories when scanned.

The first stage of document processing is scanning, which transforms paper documents into electronic images. The system uses Bell & Howell 3338 scanners and Compaq servers. OCR processing reduces the post-OCR cleanup required prior to exporting OCR'ed files.

Caere's embedded OCR technology translates the documents into a database with IntroSpect's Fulcrum search engine. The selection of OCR embedding enables document management system end users to opt for a production-level software solution. Software-enabled OCR is increasingly ubiquitous in document management solutions as it allows users to take advantage of improvements in future processing technology. So, when Intel raises the price/performance bar to faster chips, the cost of faster OCR performance will drop as well.

While OCR technology has been in existence for more than 20 years, it continues to evolve. There have been increases in speed and recognition accuracy, as well as expanding applications that use OCR technology. The need for OCR processing is highlighted by AIG's Vincent. He estimates that 70% of the Brief Base's content is received in paper format, a figure which would have been unwieldy without an effective scanning and OCR process.