Teaching Computers to Read Between The Lines

Send to friend

Businesses looking to digest the mountains of information that are locked away on paper invoices, order forms and other paper documents should take another look at AI, as it is only now living up to the hype, says document management and data capture specialist ITESOFT.

Marketing director, Tom Fox of ITESOFT UK claimed that this technology has great potential within the financial sector, as sophisticated technologies encompassing neural networks governed by business rules along with AI-based techniques are yielding positive results in transaction-oriented scenarios for financial services.

In financial systems, computers read documents in order to extract data. This data initiates workflows or is exported to upstream systems such as ERP. Recognition systems must be able to read large volumes of cheques with very little margin for error which places huge demands on the accuracy of the data capture systems but this is only possible because cheques are highly structured documents using a limited vocabulary.

However, new document and data capture solutions are trained to recognise handwriting samples of words that appear on cheques, forms, invoices, faxes and even letters and when sufficient writing styles have been learned, the recognition rates approach 100%.

Tom Fox added, "Coming iterations of this type of technology will go one step further and be able to read and then respond to all incoming mail in the very near future too."

Comments (0)

Add a Comment

This thread has been closed from taking new comments.