At Yahoo Research Berkeley we were able to extract a some notion of the "wisdom of the crowds" by analyzing the tags applied to georeferenced images on Flickr. Since the information was extracted as a side-effect of people tagging activity we were able to get an idea of what they think without requiring any work from the users - it also makes the system a little harder to game/spam.
The demo can be found at http://tagmaps.research.yahoo.com/worldexplorer.php along with a public data api at http://developer.yahoo.com/yrb/tagmaps/index.html Rahul Nair Yahoo! Research Berkeley -- http://www.rahulnair.net/blog On 9/16/07, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: > Message: 3 > Date: Sun, 16 Sep 2007 13:35:50 -0700 > From: "Charles Bolton" <[EMAIL PROTECTED]> > Subject: RE: [Geowanking] neighborhood database? > To: <[email protected]> > Message-ID: <[EMAIL PROTECTED]> > Content-Type: text/plain; charset="us-ascii" > > All, > > > > Ian is right. Neighborhood boundaries change over time and new neighborhood > definitions emerge as well. Trying to define neighborhoods based on the > inhabitants perception of boundaries is like trying to coral an amoeba. It > is interesting sociology, but it raises the question of how you update your > definitions. Maybe a scheme to let people vote on their neighborhood > boundaries and average them or display them in real time. > > > > > > _____ > > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On Behalf Of Ian White > Sent: Sunday, September 16, 2007 1:09 PM > To: [email protected] > Subject: Re: [Geowanking] neighborhood database? > > > > I'm not trying to be commercially-oriented here, just helpful... > > UMI saw this problem and built a database of informal space (which includes > neighborhoods). We have 20k+ in the US and also other parts of the world... > 'Wankers might be interested to look at our services demo: > ws.urbanmapping.com (thank you very much Andrew Turner!). Yes, it's a > commercial endeavor. > > Census tried to create a framework to allow municpalities to define block > groups around neighborhoods, but it didn't gain much traction and died. Part > of the problem is that neighborhoods are (generally) not administratively > defined. The boundaries of "Midtown, NYC?" Your guess is (almost) as good as > mine--we've licensed it to all portals, so our definition is more broadly > known. > > Because they are informally-defined, traditional notions of boundaries do > not apply. Does SoHo have to be adjacent to NoHo? Why can't they overlap? > That's how people think about space, so it follows that you can be in > multiple 'hoods at the same time. It gets messy to create a uniform > definition around something that isn't fact-based, so I don't think it's > well-suited to an open project (or should I say crowdsourcing??). > > The question also touches on a hot-button issue of mine-- the city of > Seattle selling spatial data--they've collected it at taxpayer expense, so > cost recovery is the most they should be charging, but that's not the case > some of the time. It's even worse with parcel data. The security/public > safety argument usually inserts itself here, but as many state courts have > ruled, it's hollow as a rotted birch tree. > > As an aside, the Seattle data Raj mentions might seem 'official,' but it > doesn't really matter--what people perceive as being valuable does. > _______________________________________________ Geowanking mailing list [email protected] http://lists.burri.to/mailman/listinfo/geowanking
