random thoughts and a business idea
Posted by shannonclark on November 12, 2003
or something because I’ve only been posting short stuff of late
So it has been a fairly strange past month as I have been recovering from my office move and have only just been up and running mostly fully functionally. I’ve been working out of my house at cafes many afternoons, not much different than working from home or from my old office, though I find myself making less phone calls. I need to get into a habit/routine where I make all my calls at one time and then work from cafes when what I need to do is write.
But enough on that for the moment.
More interestingly (I think at least) is a conflation of events that have been happening of late and some ideas which I have that I am considering looking into.
First, the idea in a short summary, more on it later I hope (if you are interested, have suggestions, paying investors etc, please contact me privately, if you don’t know how, look for me at Ryze and contact me there).
“A directory of the Internet that works by AUTOMATICALLY sorting pages into collections and letting you find pages that are related/similar very rapidly and quickly. Further, it would have some elements of ‘paid search’ to provide a revenue model and would apply some pretty cutting edge AI techniques for the classification and search technology”
To start it off we would be using some standard crawling techniques, but would be building up our indexes in a fairly new and innovative way. Further, rather than attempt to force pages into a vast tree ala the Open Directory Project (which is nearly impossible these days) or forcing people to get very very good at searches, I hope to have an interface and an overall data structure that would work in more intuative manners.
For example, you might be able to cut and past something into a big box on my page – we would then return results that were pages that were most likely related to the content you posted. From each page you could also then find other, related pages and/or highlight specific blocks of text (or words) that you would like us to find more pages similar to.
Unlike Google we would not have a short number of words as a limit to search phrases, nor would putting in too many words result in no results (and/or being forced to play around with word tenses, pluralities etc, over time we would address this via dictionaries, thesourses and translation abilities, eventually being able to find pages in non-English languages from an entirely English search request.
Over time as well I think this would be built into either a Bookmarklet and/or toolbar type funcitonality as well as a web base page model.
It sounds pretty complicated (and it is) but I have many years of AI technology that has been sitting on my shelf that I plan on bringing to bear on this problem – specifically I will eventually be able to pretty powerfully detect page changes, differntiate between major and minor changes (i.e. changes of actual sections vs. changes of a ad or the date), and do some other rather cool stuff to the data that we collect.
I also plan on integrating our efforts with other revenue opportunies on the web – Google AdSense, Amazon.com’s affliate programs, PayPal’s programs and the like (perhaps Overture though I think Google might make more sense for us).
As well, I will likely offer “sponsorships” of certain things, though exactly how that will work is something I have to figure out since unlike an Open Directory type project we won’t have very fixed categories, though we may have something close enough for sales purposes.
My underlying technology likely would cause some developers to have fits – I don’t plan on doing this in a very traditional manner… but heck that’s probably why it will work.🙂
In other news, I have sparked a bit of a fit over at ChowHound with a post I made about Spoon Thai and it was followed by another discussion on the “Not about Food” Chowhound board on Secret menues. Saturday I may be joining them for a tasting of Italian Beef around town, not sure exactly what the reception will be, but personally I think the discussion while at times getting too personal has been interesting and useful and entertaining.
Last night I finally saw Memento on DVD, a movie I really should have seen when it was out in regular release. I originally read the story first published in Esquire Magazine so I had some idea what was going on, but still it was a very enjoyable puzzle film and after watching it, like any good net person, I went looking for and read some online discussions of what “really” was going on (as well as tried to get the supposed easter eggs such as showing the film in time order to work but could not).