Using Unstructured Content to Update & Audit Process Management Systems
Data elements entered in enterprise process management systems are often taken from and evidenced by supporting documents. For example, in home mortgage loan tracking systems, names of borrowers may initially be taken from original loan applications. As more documents are generated during the loan process the borrowers’ names and signatures will also appear on the additional documents and are hopefully consistent.
All stakeholders, including managers, shareholders, and regulators, depend on management systems accurately reflecting the data shown on supporting documentation. Being able to detect variances enables the enterprise to guard against deliberate or inadvertent misstatement of asset values and risks.
When all documents involved in transactions are visually classified and assigned document-type labels, the data elements or document attributes that typically occur in each document type can be mapped to the database tables and fields that drive the process management systems (see graphic).
The mapping serves two purposes:
- Data Input. Rather than having to constantly retype the same information, data elements can be automatically extracted from designated places in specific document types, consistently formatted, and placed in the database. This speeds processing, lowers input costs, and enables automatic verification of data values being added.
- Auditing/Validating Structured Content Data Values. Automatic file classification and document attribute extraction permit automated audits to confirm that values in control systems accurately reflect data values shown on supporting documents and that those values are consistent across all documentation. Signatures from multiple documents can also be presented in a way that speeds comparison for auditing/fraud detection.
For a description of how visual classification works, see Technology.
For use cases in specific industries, see Use Cases.