Content area
Full text
Despite the hype concerning the move to paperless systems, the bulk of mortgages are issued on paper. Because of this, the need for data extraction from paper documents during loan quality analytics arises from the relentless drift that occurs between loan origination systems and the legally-binding papers in a mortgage.
A core tenant of the systems and processes dedicated to ensuring loan quality is that the data in repositories are fundamentally untrustworthy. It's estimated that anywhere from 10% to 30% of the information in a typical LOS is inaccurate. Systems dedicated to ensuring loan quality need to use data from a variety of sources, performing reconciliations and deficiency analytics based on multiple sources. One principle source of data is the paper documents. It is the paper that constitutes the ultimate source of "loan truth," since these documents are the legally binding obligation of the loan. In addition, in many cases, the original LOS data is not available for review, leaving only the documents.
Enterprise data entry applications are required to transform the information trapped in paper into usable digital data. Such systems are surprisingly complex, embodying workflow rules, mixtures of automated and manual data extraction and management of vendors and operators in multiple time zones. In addition, security agreements, reporting and service-level agreements on data accuracy are also crucial. As only the first step in a comprehensive process of mortgage analysis, data entry systems must also integrate with other processing applications.
Data extraction occurs in a stepwise fashion according to a well-defined linear workflow. The workflow consists of three distinct stages, each of which can be broken down into multiple sub-stages. The first stage is digitization of paper documents using scanning. The second consists of cataloging, or indexing, each of the pages of the scanned documents. The third stage is extracting key data fields from the documents.
Scanning is a well-established technology with a huge...