Continuing this week’s them of rejection (a bloke could get paranoid), I’ve just heard that my (more in hope than expectation) 4iP project has been knocked back. If you don’t know what 4iP is then a quick look at their website should help, if you don’t care then it doesn’t matter (basically, funding for “public service interwebs”). I wasn’t after the money so much as the promise that they could link you up with other people that would complement your skills —I couldn’t have made my idea on my own.
My thing was very basically a “geo-stumble upon” a way to attache geo-data to the wealth of exsiting web content that doesn’t have it. Easy to use, but mainly building up a huge bak of useful data that people could do cool stuff with. Not local search, but “geo attention“.
Here’s what I submitted, in case you are interested — I think this would be a very good idea, so in lieu of me being involved in making it I’d like to throw it out to anyone who might:
Geo-data is about to be the next big thing on the web, but people who make sites don’t know how and where sites are useful. People are also to busy to tag thier content. A bowser plugin/web service that allowed people to say “this content is useful to me here” or “this site is about here” — and then used this data to place sites on the earth as well as on the web (or at least peoples’ attention to them). Site plus open API allows local search for existing content — creating geo-data for the millions of sites without it.
There are billions of pieces of content on the internet, but very few have any geodata attached to them — some because they aren’t geographically relevant, but many because the will/skills aren’t there to attach that data. One of the most difficult tasks on the web is local search, and even if new content is geotagged the existing net isn’t — this will really become apparent with the “hyperlocal revolution” having huge gaps.
Technically the idea is to make recording your “geo-attention” of a site as simple as possible — just one button click when signed in (using existing services: fireeagle/skyhook/google latitude etc to record location). Your profile can then be used to recommend sites based on your interests and location, you share your attention data, “friends” share your geo-bookmarks.
I’ve been experimenting with my.loki.com, which aims to automatically update your position in the world (and it’s scarily accurate), it will also share that information with other web services if you allow it. FireEagle from Yahoo is to offer the sharing too, if not the automatic stuff. It’s helped me think around a topic that’s been sitting in my brain – I’ve been thinking about how to model our relationship with notions of place and identity in a way that web services can use. The mark-up itself is not something I’ve been worried about just yet, but I’ve been thinking about the sheer amount of data we need.
GPS on our smartphones will no doubt be able to provide up-to-the-minute geo-data of our actual positions to web-services in the near future, by there are any number of other places that will inform our areas of interest in news or events. In the same way that attention data will build and share what we’re interested in, we’ll need to build geo-attention data – of our physical places of interest. Some may be gathered by the same means, based on the attention we pay to geo-located items of news, but some we’ll have to set:
Where I Live – I set this now, on Facebook at least with the neighbourhoods app.
My Region – the wider city or county. This you would think would be a simple one, but administrative boundaries bear very little relation to the ones that exist in our head. Witness the endless debates about what’s ‘The Black Country’ or what is or isn’t in Birmingham. Different bodies also have wildly different views on what is contained in a region (an example int he comments here).
My Country – An easy extrapolation, in most cases, border disputes notwithstanding.
My home town/country – I live in the city/country I was born, but if I didn’t there would still be types of news I would want from them.
Where I am – a job for loki, or FireEagle or their successors.
Where I’m going – there are certain things you want to know about the future, social travel sites or your calendar may offer this data, if we can get it out.
Where my friends are –or any other relationship you may have with a place. This can only really be set by you, but a mechanism to do it is needed.
While geo-tagging news or events is developing on point data (and may need to expand this in some way to be useful), many of our geo-attention centres are even fuzzier. Geo-attention profiling based on these news items will be crucial, but there are other things needed. Geo-attention data needs to start with either squares of long/lat or even HTML image-map style regions, and then fuzz them – creating “geo-attention heat-maps”.
There are still areas that require our input – I’d be willing to set them if I got me the targeted content that saved me time and I’m sure others would too. The data is so complex that is needs to be open, transferable and easy to apply, but it may be some of the most important knowledge of our habits that can be used to improve the quality of information we get.
For Geotagging functionality to be anything more than a cool widget, it must also…
* Let users to define and save multiple areas of interest.
* Support physically defined regions.
* Account for more than just the hyper-local or the global scope.
* Incorporate proven organization schemes like topical categorization.
* Include news that is tagged to multiple locations, irrelevant locations, or no location at all.
* Feature a truly great interface that supports everything on this list.
I’d agree most of this - especially with idea of multiple places.
Although on the idea of regions, I’m thinking that using longitude/latitudes with less detail (eg 52, -1 instead of 52.122172, -1.23233 &c.) can do a lot of this without thinking about the geotag equivalent of HTML image mapping – which would be unnecessarily complicated.
The geo information isn’t going to be the only data used in deciding the interest of the user – AMPL or some other attention data will be factored in too. Most news does happen at a specific point, and will affect greater areas due to its importance and relevance, rather than a drawn physical border.
I’ve just let other people see my new project up your end, (amazing what popping a url in your facebook status can do) which is a little toy for geographically mapping Birmingham things on the interweb. It uses geotagging, and here’s a little explanation of how it works.
Although some sites, such as Flickr, will pump out geotagged feeds they aren’t necessarily in the correct format for overlaying on a Google map (there are three competing geotag XML formats for a start). Luckily, the Location Extractor operator in Yahoo Pipes will sort that out, as well as generating geo information from posts in feeds that don’t expressly geotag (upcoming.org’s feeds are a good one for this as venues have addresses). While you’re Yahoo Piping, you mas as well filter in some other ways: I restrict some of the longer feeds to ten posts, and the Flickr pictures to those that contain a Latitude of 52 (Birmingham in the UK is at 52°N, the many other Birminghams aren’t).
You can dispense with the whole fiddle of Piping your feeds if you are creating them of course, and you can geotag items accurately without having to oddly list the full address of what you’re talking about. Birmingham: It’s Not Shit is based on WordPress, and as such there are plugins to do the job for you. I’m using Geo and GeoMashup (which will generate Google Maps with your posts on with just a quick inline tag, see BiNS). I first tried the seemingly more powerful GeoPress plugin (also available for Moveable Type blogs), but despite working well as a tool to use it wasn’t generating valid XML for me.
Geo places Latitude and Longitude boxes just below your post editor, but also allows you to store locations and select them from a drop-down menu. Locations are stored in the plugin options page (which will also set a default location – your house? The centre of town? – for posts you don’t expressly tag):
The only problem with this is that until you build up a database you’ll be spending ages finding the geolocation for each post – and there aren’t really any simple web-tools that do it for you.
Although, GeoMashup will place a handy Google map on your post editing page, as well as a Find location box – you’ll have to click on the map to reveal the lat and long for the position you find and then copy and paste them into the Location fields higher up.
Fill your stored location database, you’ll need it!
Pop over to Google Maps and get yourself an API key, you’ll also find plenty of example code. View Source at this page to see the bare bones of the code up your end uses. For overlaying RSS feed information it’s quite simple:
Use GGeoVml to load the feed:
var geoXml = new GGeoXml("http://www.birminghamitsnotshit.co.uk/feed/");
And then add.Overlay to place it on the map as you draw it:
You can add as many overlays as you like, although it is slow to draw too many – that’s why the toggle buttons are handy. You also waste processing power and time by placing markers off the viewable map – that’s why filtering the feeds was useful earlier.
I’m not sure how useful it is at the moment, adding all the feeds at once is a little slow and the Pipes are sometimes flaky. In fact it doesn’t have a great deal of practical use at all. It’s still an interesting visualisation, though and I feel that the real killer application for geotagging is about to hit us – so anything you do to make your work tagged correctly will give you a leg-up as soon as it hits.
If you’ve got a Birmingham based feed and geotag the entries, let me know (email up the right) and I’ll add it to the map.