aiSTART

When a word search is not enough, you need aiSTART

Your document management system can already search for words and phrases in a document. With today's search engines, that's pretty easy. But can your system find the information you want to put into a database when all the documents are unstructured (or free-form)? If your data entry staff is reading through complex and variable documents looking for the correct information to enter, you need aiSTART.

The aiSTART workflow begins by reading each page using state-of-the-art optical character recognition (OCR) algorithms. Then its Advanced Learning Engine, using a pre-primed Knowledge Base, is unleashed to seek out the information which belongs in your database. Finally, a feedback loop assures that the process continues to improve its accuracy.

Training the system is simple and quick. Without software coding, it may be set up by a business analyst. In most cases only minimal industry knowledge and document understanding is needed.

The internal algorithms to extract information of interest may either be inferred from the existing experience of historic data entry or designed interactively. Once implemented, a feedback mechanism enables an application to improve in data capture accuracy over time.

The computer-intensive back-office aiSTART processes have been expertly engineered to provide reliability and scalability. aiSTART easily integrates with existing or new data capture applications and work flows, providing dramatic increases in accuracy, efficiency and throughput.

The technology has been honed in the real world through the development and deployment of two successful commercial applications: aiINDEX and aiREDACT. The case studies provide additional insight these products.

© 2009 Mentis Technology, LLC.