From: Aldric Giacomoni Date: 2010-04-02T03:00:49+09:00 Subject: Re: Playing Games with "Ruids" Seebs wrote: > On 2010-04-01, Andrea Dallera wrote: >> Please everyone stop feeding the troll (or whatever he is). >> I appreciate the effort but please just leave him alone, he'll run out >> of gas soon. > > I think you're right. > > I'm done; he's established to my satisfaction that he is either trolling > or fundamentally incapable of communication. I hope that, if there was > a > good idea in there, it eventually occurs to someone who can use > language. > > -s Hey Seebs, I've got this great idea. It's going to involve a kind of giant database, maybe nosql, maybe relational. It's gonna have a ton of information. In fact, it's going to learn to analyze data without us telling it what to look for [1]! And then we can plug in more information about the data. That's going to create links within the data, which we can probably use with neural network back-propagation to create and refine existing filters. Now, about those filters - they can be like agents, who watch changes to data and fire off an email when specific data changes. See, for instance, we have this database with lots of pharmaceutical information, right? Lots of tables, lots of data, lots of research. Now, we're also going to have a database for patients which links to information in this huge database of pharmaceuticals (which I will call pharms from now on). So, for instance, patient John Doe is currently on this pharm called foobarbazquuxelamine [2], and it turns out that new research indicates that the drug was found to cause cancer in laboratory mice [3]. So, the database is updated, possibly with a red flag. The referring physician / lawyer / surgeon / patient has this alert set up for the patient record. If it were me, I'd have alerts set up for all my patients, but moving on. This alert sends him an email, "Oh, #{deity}! #{this patient} will get #{cancer} because of #{this pharm}!" Magic. Except not. Alright, so now we just have to implement that; we should just create a DSL so everyone can implement their own agents / filters / database queries. The client program will just start as many of those queries as we tell it to; how about 2,000 per second? Should be fine, right? I wonder what'll happen when one million physicians across the world each fire two thousand requests per second. [1] I admit it. I am a dreamer. [2] It was invented at MIT, dontchaknow. [3] Everything causes cancer in laboratory mice. Life has been found to cause cancer in laboratory mice. Mice have been found to... Well, you get it. -- Posted via http://www.ruby-forum.com/.