Tuesday, August 09, 2005

Artificial Intelligence

Yesterday I travelled to the Registry of Deeds in Worcester to see a presentation on a computer program that would automatically index documents. The developer of this program, called AIindex is Mentis Technology, LLC and their website is http://www.aiindex.com/. The program seems to use a super sophisticated OCR (optical character recognition) technology. For those of you not familiar with OCR, we've been using it at the registry for administrative purposes for seven or eight years. (we've never used it in any way with recorded documents). Here's an example of how we might use it: A consultant gives us a report on paper but not in an electronic format. We want to use large portions of the text of that report in a new document. One solution would be to retype the portions you want. With OCR, you scan the original report, creating an electronic image of the document, then the OCR program "rescans" this electronic image, converting images of individual letters to word processing characters. The earliest versions of OCR software, yielded some pretty bizzare results, with "O" coming out as "Q" or "8" so you had to do quite a bit of editting. But OCR software is much more sophisticated now and AIindex seems like its the state of the art. There are many potential uses for this. In other parts of the country, registries are scanning newly recorded documents and allowing the OCR software to do the indexing. For example, the software would scroll through the document image looking for the words "grant to" followed by something formatted like one or more names. Those names would be placed in the grantee fields in the index. All fields would be populated this way. Then a registry employee would simply verify the data that was entered by comparing the names and addresses in the data fields with those in the document. It's supposed to be a much faster process than typing all of the information as we do now. With our need to get names into the index as soon as the document is recorded, I'm not sure this process would replace recording counter data entry in Massachusetts (the document must be scanned and then the software goes to work which all takes time - although not that much time). Still, the software looks like a fascinating tool that gave me a glimpse into the future. It's so good, that we should probably get it and then figure out how to best use it. But that's not all, this software has another very interesting application, but I'll write about that on Thursday.

No comments: