Archive for December, 2008

Flickr Machine Tags for museum objects

Published
by
Giv P
on December 18, 2008
in General and Searching
. 1 Comment

At the museum we have been experimenting with machine tags and allowing the user to add their own photos to collection object pages. We have successfully implemented the Flickr API to display images of objects that users have tagged with our machine tags. But rather than just using an arbitrary string, Richard Barrett-Small from the V&A suggested that we come up with a convention that all cultural institutions can use. We have used the following format:

culture:[partner name]=[object ID]

So for the V&A, for example, to pull up images for an object with ID 12345 we would use culture:vam=12345 and an object at the TATE could use culture:tate=abc123

We thought it would be good to put this idea out there and see what others think. The advantage of having a convention like this would mean you can potentially predict the machine tags for each partner and different partners can embed/share images using a portion of the tag. For example, if I wanted to extract all Flickr images at the TATE I would look for images tagged with: culture:tate and I would leave out the ID to extract all images like a wild card.

Not very sophisticated. Ideally you want to be able to add more machine tags like categories so you can query all V&A images in the “painting” category. But since it will be the users who will be tagging these photos, the process should be as simple as possible.

I would love to hear from other museum people about this.

On Google Friend Connect

Published
by
Giv P
on December 8, 2008
in General
. 0 Comments

Last week I posted something on Facebook Connect and over the weekend I managed to play around with Google’s Friend Connect. There’s already been a lot written about the two so I won’t bore you with yet another Google vs Facebook post. But what I will say is that as a developer I found it much easier to integrate Facebook Connect into an existing application even thought it took longer to implement.

With Google Friend Connect I was up and running in less than a minute by copying and pasting some Javascript code into my site and all of a sudden I had a friends list, comments and ratings widgets without any coding at all. But that’s about as far as I got. I haven’t figured out if it’s possible to pull out info from the widgets through an API. It seem like you can quickly add a bunch of social widgets to your site but if you decide to fully integrate everything it will be very difficult to do.

Over all, I think they are both on the right track but neither are quite there yet.

Image Colour analysis API

Published
by
Giv P
on December 4, 2008
in General
. 2 Comments

Earlier this year I posted something on image recognition and colour analysis of museum objects. In the spirit of sharing, I’ve decided to create a simple API for people to use.

Send the API the URL of an image (only JPGs at this stage) and it will return a list of colours used in the image in an XML format. The values are red, green, blue and hex. The API will dither your image and will grab the top 50 colours used in the image.

Usage: http://www.givp.org/projects/colour/?img=[encoded jpg url]

e.g.
http://www.givp.org/projects/colour/?img=http://www.desktoprating.com/wallpapers/animal-wallpapers-pictures/4-cute-puppies-wallpaper-640×480.jpg

And obviously, please play nice :) tell me if you’re planning to analyse lots of images.

Enjoy.

Facebook Connect & Museums

Published
by
Giv P
on December 3, 2008
in Experiments
. 2 Comments

I have mixed feelings about Facebook Connect, but let’s talk about the pros first.

In case you haven’t heard, Facebook officially released their Connect feature this week and it allows any external site to take advantage of FB’s authentication system and full API. This was only possible through FB applications in the past but now you can open up your own site to over 120 million active FB users. They can log in without having to go through a registration process and you will be able to pull-in almost every information available on their FB account such as email address, avatar, status message, photos and a list of all their friends.

The potentials are obvious. But how easy/hard is it to apply this to an existing application on your site and how useful is it to the museum community? So I got my hands dirty and dug straight into the documentation. I used an existing framework (Symfony) that I have used to develop a collections application. The challenge was to first integrate Facebook Connect on top of Symfony’s own authentication system and the second challenge was to use FB’s API to extract information that I can use on a museum collection website.

And voila! it took 10 mins to set up the authentication system and another 20 mins to do the rest. FB Connect is ridiculously easy to implement and once you’re authenticated, you can use their PHP classes to make calls to their API and get virtually any information you want about the user.

Here’s my information on a museum collections site after I have been authenticated

FB Connect

The possibilities are endless. You have the user’s email address, so for example, if they wanted to purchase an image, you could take them through an e-commerce system or do something fun like let them add the object as a favourite and publish the image and object details on their Facebook profile or update their status automatically.

What makes me uncomfortable about FB Connect is that you are depending on a single site. It’s not like OpenId and it’s hard to know whether FB will be relevant in 5 years time. So it may be best to keep your own sign-up process and add FB on top as a feature.

The other factor that makes me worry is that you don’t even have to go to Facebook to log in, unlike OpenId. You are presented with a DHTML login window directly on your own site. So what’s stopping people from making a bogus login window and storing your login details? Surely I’ve missed something there.

But over all, I have to admit I am really impressed :)

Update:

From FB’s official press release:

Over 100 Websites implemented Facebook Connect during the testing period or are set to release in the coming weeks. Examples include: Citysearch, CNET and TheInsider.com by CBS Interactive, CNN.com/Forum, ConnectedWeddings, Gawker, Global Grind, Govit, Howcast, IndieGoGo, Inside Facebook, Joost, MoveOn, MyBarackObama, Newsbrane, Red Bull, SGN iFun, SFGate, TechCrunch, TripAdvisor, Yammer, Vimeo, vLane, and Xobni; alumni associations at Oklahoma State University, University of Toronto, Oregon State, and Ithaca College; as well as services and plug-ins from Force.com, iModules, Pluck SiteLife, and Six Apart.