Why Free Form?

No Environment Can Predict The Layout of Every Paper Document it Receives

Capture and recognition systems have traditionally worked by using templates. The recognition system is "taught" about a series of possible layouts of a variety of document types. When a document with that layout is seen, it is recognized as the correct type and data is extracted.

The problem with this model is that when a new document type is received, it initially is not classified at all. Manual intervention is required, initially to manually classify the first few of these documents and then later the system must be taught about the new document type. A similar scenario results if the layout of a "known" document changes.

The net result is that recognition rates are fine immediately after a "training" period however they deteriorate over time as the profile of incoming documents changes.

Why Open?

No One Should be Too Dependent on External Consultants

Proprietary recognition systems require the intervention of a third party such as an integrator or the software vendor itself to initially create templates for document recognition and then intervention to "tweak" the system from time to time to maintain desired recognition rates. From a customer's perspective this can end up being inflexible and expensive. The person that intimately understands the profile of incoming documents is the end user. Open recognition systems allow the end user to create a recognition system that matches their own incoming document profile. The end user can also modify the system later if this is required.

Enter Prosar-Aida ...

Prosar-Aida is a free form recognition engine that uses artificial intelligence (AI) to determine the type of a document and then the locations of the data items to be extracted. Prosar Aida is taught about the general properties of all invoices (say) and is able to identify and extract data from invoice types it has never seen.

Prosar-Aida comes with an easy to use graphical user interface that allows end users to teach it the general properties of all major documents received. The end user also uses this same user interface to teach Prosar-Aida about the general properties of the Total Invoice Amount of an invoice (say). Prosar-Aida is able to apply these general rules to invoices that are already well-known as well as those that the end user might receive years from now.

The benefits should be clear. Prosar Aida is an open, user-driven engine that detects documents whether they are known or not.

Case Studies

German Health Insurance Company

Deutscher Ring Hamburg Group

Horticultural Trade Association

Press Coverage

Information Week

Documentation

See our download page for further information on Prosar-Aida^TM, including a download of a product white paper and brochure.