[With apologies to Rod Stewart] You’ve heard the saying that a picture is worth a thousand words. Twitter, the new online social phenomenon, lets people tell their story – 140 characters at a time. But there are no pictures.
Now, there’s been a healthy debate about Twitter here at 901am. Duncan asked does Twitter have substance? David told us why he Twitters, Muhammad gave us a contrarian view, and Sharon suggested the top 5 ways smart people use Twitter. Twitter won’t replace blogs, yet it’s market share is up 55%.
Although I tried Twitter out, for me it’s more interesting to follow what others are saying than to offer my thoughts 140 characters at a time. And being the ultra geek that I am, I have more fun analyzing the content of social spaces than actually participating.
So I thought I’d mashup a Twitter feed and play with it. My first thought was to use a Yahoo Pipes pipe to replace the tiny amount of text in each Twitter message with a flickr image. Punk legend Henry Rollins‘ Twitter feed was my first try. While I like reading his feed, Pipes’ content analysis module didn’t produce enough variety in his Twittering to generate different flickr images. I tried Nick Wilson‘s Twitter feed (he who was a former co-owner of Performancing). Same problem.
It’s not that either is boring, just that Pipes’ content analysis and the available flickr images didn’t jibe with either’s Twittering. To get more variety, I used Twitter’s recent public updates, and voila, it worked. A very different type of time waster: a slice of flickr driven by Twittering, fueled by Yahoo Pipes: Every Twitterer Tells A Story (potentially NSFW). Though I couldn’t view it from my custom Teqlo feed reader.
Of course, the resulting flickr partition might be more interesting if the original Twitter message text were included. However, the Yahoo Pipes annotate module does not work properly when you try to view the images outside of Pipes builder interface.
Originally posted on March 23, 2007 @ 9:41 am
raj says
If you try the Pipe out, don’t forget to click on the “Run this pipe” link on that page.
Broc says
i do not understand this at all
i guess ive got to go fool around with yahoo pipes and get a general sense of whats going on
raj says
Broc: Yahoo Pipes basically lets you take one or more web feeds and filter them using a few text-matching rules. That way, you can customize a group of feeds to only give you, say, articles that have the phrase “new media” in them. There’s more to Pipes than that, but that’s the gist of it.
raj says
Upon revisting the problem, I figured out how to used Henry Rollins feed to generate images from flickr. Though now I realize that what I’m saying above may not be clear. What I’m essentially doing with Yahoo Pipes is taking a Twitter feed and replacing each message item with a corresponding picture from flickr. How do I know what pic to use? I use Pipes’ content analysis module to scan each Twitter message and decide on a keyphrase, which is then used to retrieve the first flickr pic found with a tag similar to that keyphrase.
So in essence, if you pic a single Twitter feed, running the Yahoo Pipe is like producing a life in pictures for that person. Here’s Rollin’s life.
in pictures. If you don’t see anything, click “Run this pipe” a few times. Yahoo Pipes is still buggy.
qDot says
Hi! I’ve just gotten the same idea up and running (found yours through googling “flickr twitter mashup” ’cause I figured there had to be a few people with this idea :) ), ‘cept I used python and some natural language processing. Twitter + nlp is a rather hard problem though, thanks to bad grammar, tiny sample space, and the non-existent corpus needed to train for internet speak and up to the nano-second pop culture. ^_^
Still not exactly perfect, but getting there, and a lot less blackboxy than Pipes.
http://www.30helensagree.com/fwiktr/
raj says
qDot: 30 helens agree? You must be a “Kids in the Hall” fan, despite living in the USA.
Good luck with fwiktr.
I love Yahoo Pipes because it lets me rapidly protoype. In some cases, i’ll also never need to write custom production code.