OPML Sampling using Yahoo API

I'm slowly but surely coming around to the benefits of OPML, which stands for Outline Processor Markup Language. Developed by Dave Winer, OPML is being used more and more as a compliment to RSS. A common use for OPML is to import and export one's RSS subscriptions into or out of an RSS Aggregator. Another application of OPML is a Reading List - which will enable you to subscribe to a group of RSS feeds and whenever the group is updated, the Reading List automatically updates too.

Today I saw another intriguing use for OPML. J Wynia has been playing with OPML and has come up with a neat concept he calls OPML Sampling. What he did was take an existing OPML file, in this case from the Web 2.0 Workgroup (disclaimer: I co-founded it and this blog is in it). The Workgroup has 20 feeds in its OPML file and J used the Yahoo Search API in order to carry out the sampling. He described it here:

"We’ll take an OPML file, load a snapshot of each feed’s current contents. Then, we’ll loop through each feed item and ask Yahoo how many pages link to it. For each feed, we’ll extract the one item that is most linked to (aka the equiv of Google’s pagerank for choosing the top page for a search). That “top” post will be put on our sampler page as the example of of the feed’s content. The sample page will then contain one item from each feed as well as some information about the feed itself."

The result is this page, which has one post each from the 20 Web 2.0 Workgroup feeds. Alex Barnett calls it a "a dymanic *best of*". Mike Arrington, a co-founder of the Web 2.0 Workgroup, is also intrigued by the OPML sampling - but wants to make it "more real time". J Wynia responded in the comments to Mike's post by saying he sees the OPML sample as a measure of 'interestingness' (I'm borrowing the Flickr phrase).

The OPML Sample by J Wynia is a fantastic example of blending the mash-up world of APIs and the content syndication world of RSS and OPML. I'm looking forward to experimenting with OPML myself and I hope J does more samples like this. Incidentally, if you want to read up on OPML then this post by Kosso gives some background. Plus Dave Winer writes a lot about OPML these days, so there's lots to learn.