Tag: Information Science

Spotify Recommendation Engine

Loving the Spotify recommendation emails I get, although probably not for the reasons they’d prefer. The best thing about them is that they give the reasoning behind the recommendations. Based on this you can get a rough visualization of the Eigenvalue matrix they use to make these suggestions. For example:

Simon & Garfunkel

The most successful folk-rock duo of the 1960s, Paul Simon and Art Garfunkel crafted a series of memorable hit albums and singles featuring their… (Read more)

Because you’ve listened to Buddy Holly, Jimi Hendrix, Johnny Cash and The Jimi Hendrix Experience.

This is probably the most novel suggestion I’ve been given. Not novel in the sense that I’d probably never listened to Simon & Garfunkel, but novel in the way the suggestion came about. None of the vectors (groups/bands/artists) used to make the suggestion are close to the folk-rock genre yet if you average out the vectors, I can see how you’d land squarely along the vector that gives you Simon & Garfunkel.

Another interesting bit of vector math is the one that came up with The Clash as a suggestion. Here we have:

The Clash

The Sex Pistols may have been the first British punk rock band, but the Clash were the definitive British punk rockers. Where the Pistols were… (Read more)

Because you’ve listened to Beastie Boys, Jimi Hendrix, The Jimi Hendrix Experience and The Kinks.

Again, you can see where The Clash are like the pre-rap Beastie Boys with more complex songcraft than most “punk” bands a la Hendrix, with sometimes clever lyricism a la The Kinks. This recommendation also makes sense, even if it is for a group that is not all that far off the beaten path.

Vector analysis and the related mathematics was probably the one subject that completely blew my mind back in grad school. Recommendation engines are truly fascinating things. Not only do they possibly show you some synthetic reasoning, but they also give you some pretty decent insight into the folks who set up the matrix that is being operated on. So even though the recommendations Spotify sends along are pretty middle of the road, they still provide some great entertainment.

The Code4Lib Journal

The code4lib folks have launched their e-journal! I first heard about this project, and the group itself for that matter, at ASIS&T’s 2007 Annual Meeting. Although the whole conference was beyond excellent, code4lib was one of the top five highlights for me. Enough of my blather though, here’s the link:
The Code4Lib Journal

I found the article on the development of an OPAC API to be most interesting.

The Brilliant Idea

So I’m sitting at the desk whiling away the hours until midnight by reading blogs from back when I first discovered the activity. Hats off to LukeLog and Meg (who used to be NotSoSoft back in the day). Reading as these kids / folks / people really invested a lot of effort in exploring the form. I certainly won’t get in to who discovered what when, but for me these folks were pioneering. It was through them that I discovered MeFi which actually was side linked from the Underground IIRC.

What really strikes me now is how conversant they were with the form. Not necessarily the most technically / design / content incline but definitely the best at reviving the vanity site of yore. For me they hit the right balance between geek, personal life, and connections. It was all quite inspiring as I was just starting to rediscover the web myself after years of no computer and no connection. I was testing the waters with graduate school and information science and I saw how I could integrate what they’re doing with my own ideas about information presentation and retrieval.

At the same time I was also working with/for a guy who was strictly old school in the technology department. Not that he didn’t have nice toys or shunned new technology (although he does the latter somewhat) but he was rigid about the least tool necessary for a task. We did not use a graphical IDE for development. We definitely shunned the use of flash and minimized the images. We were on the edge w/r/t implementing CSS driven sites. Server-side scripting? Rriiiiigggghhht.

I’ve taken a lot of that forward with me. In my MIS program I was a bit snobbish about the HTML and code my classmates generated. I often let this snobbery get in the way of appreciating the larger lesson. Moving into my first real, bona fide professional position I made my life a great deal more difficult because of some zealous adherence to an amalgamation of XHTML Strict / CSS / 508c / Nielsen rules. Now at my most recent position I am still fighting my first boss’s battles.

And I’m really not sure it’s worth it. Either that or I’ve ended up in the wrong place professionally. Slowly but surely, the web design field is moving beyond my skill set. I’m very focused on programmatic solutions and automated text processing while the broader audience could care less about these things. I keep waiting for the vindication. Well, okay, the CSS / XHTML thing is starting to pay off. But I’ve got staff who think nothing of bundling instructional tutorials in Flash instead of interactive DHTML. I’m falling behind playing with AJAX because I continue to have to keep the bailing wire web site I inherited together. And I don’t have the opportunity to pursue this on my own time because I’ve elected to try having a social life and meaningful relationship outside of work.

Then there’s the whole missing out on ontologies and metadata angle which was the thing that most interested me in school. I just cannot seem to find an entry into the professional ranks and now I’m not even certain what’s left of my knowledge is even relevant.

And so I whine as I sit at the desk whiling away the hours until midnight…