Friday, April 15, 2011

Intelligent Document Management: One Use Case, Putting all together

In this video we provide an example that sums it up for what intelligent document management is all about:

0. We upload a scanned document (a non searchable PDF, just an image). 

1. We extract text using OCR technology, and then w
e extract (using our semantic reasoning engine) relevant data (tags), which are keywords inside the document or just concepts related to the document which may not appear in the document but is useful to describe it. We do this using DBPedia, OpenCalais and other semantic web cloud environments.

2. The document type is recognized: contract, and in that type, the exact sub-type of contract. This feature can also be used with other enterprise document types, such as invoices, records etc. 

3. Based on the above, a workflow involving workers in the organization is started. The system can be configured to take in mind things like staff workload etc. to asign the reviewing/approval workflows to whoever is less busy and may perform the required task from a rights/operational point of view. For example, an invoice will trigger an "invoice pay/approve workflow" and documents will be routed to the persons that perform this actions, depending on data inside the customer such as "customer" etc.

4. We perform several (3) searches using the terms inside the image (full text search). This terms are like the EIN number (company number) and the ID Number of a person. 

5. We show the graph of the tags extracted on the document (relations graph).

There's plenty more videos, with automated invoice processing (using semantics, not just zonal OCR) etc. and we have this working in production in Spain's DMV, the Spanish Army, and large financial institutions (BNP Paribas, BBVA, Cajasol etc.)

This video summarizes perfectly a complete case of use of the entity-tag extraction.
This can also be done with a mobile client (we have developed one for Android-iPhone which is available here) and using then Geo Local tags to do cool things like finding documents which have been uploaded to the system by users close to me or by users in one particular region.

We can also improve this use case taking a picture with the phone, one click upload from the invoice and then the whole thing described before (1 to 5.) will take place, also like I said, with the geo-tag include. 

LikeUs Yerbabuena Software on LinkedIn Share

No comments:

Post a Comment