Amazon.com, the internet retailer, launched a set of web services way back in 2002. For those unfamiliar with the term 'web service', it generally refers to a program running on a computer somewhere on the internet that you ask to perform some task, then return the result to you over the internet as well. So, it's just like it sounds, a service performed by a computer over the web...
I heard about the launch of Amazon's web service, and have heard smatterings of talk since then when a new service is launched, but never paid much attention until I saw their payment service. So, with this service, you create an account with amazon and pass them payment information over the web and they send a response back to you indicating the status of that payment. It's like a merchant account for credit card processing.
Anywho, I had an idea a few weeks ago that the family think tank didn't quite understand. Taylor understood what I was trying to get at, but I wasn't explaining my idea in terms the family could understand. The idea was basically human, distributed OCR (optical character recognition - the process of turning a computerized image into text).
I would write a program to break an image (a scanned document) into words or lines and present those randomly to humans on the internet. Each person would be presented with a small portion of that image that contains a word or sentence and would type what they read in a box below it. If after a certain number of people have seen that portion of the image, and agree that it says the same thing that section of the image would be assigned that value. After the image's pieces have all been converted and agreed on, I could piece the document back together and have a text value for the document.
Today, after looking around at their web service offerings, I found a service called Amazon Mechanical Turk, which allows you to break down tasks and submit them to Amazon, who will in turn assign them to a global group of workers. One of the companies using that service offers an optical character recognition service to turn document images back into text.
I'm still trying to find whether there's any way to turn some of my ideas into feasible options with services like these. Two of the ideas I ran past the family within the last fews months seem more attainable now, but still not necessarily workable. I don't plan on quitting my day job as a programmer (or my night job as a lazy husband), but like to toy around with spare time money makers.
Posted by Jordan at February 16, 2008 2:05 PM | TrackBack