Hot diggity

The Collective Intelligence Foo Camp looks awesome, especially if you're into blogs and recommender systems (and who isn't, right? Right...? Everybody?).

Some particularly cool attendees:

• Greg Linden (worked on Amazon's recommender system, great blog)
  
• Matt Hurst (ex of Nielsen BlogPulse, now at Microsoft, great blog)
  
• Niall Kennedy (ex of Technorati, Microsoft, bunch of other places, great blog)
  
• Bernardo Huberman (also talking at ICWSM '08 next month, sad lack of blog)
  

(via Greg Linden)

Uh, lazyweb...?

Update: Thanks, universally helpful lazyweb!

I know it was only a month ago that you were invoked here last. But... Freebase. It's killing me (I was inspried to finally build something using it by Pierre, whose examples I have sadly failed to adapt :( ).

What MQL can I use to get *all* of the information about *all* people with profession of 'author'?

I have

{
  "query" : [
  {
    "*" : [
       {}
    ],   
    "guid" : null,
    "limit" : 5,
    "name" : null,
    "profession" : "author",
    "type" : "/people/person"
  }
  ]
}

(thanks to Pierre for pointing out a syntax error here originally, oops)

Which should list all of the properties for 5 authors, right? But I only see properties from the /people/person schema.

How do I get all the author properties too?

I ended up using the MQL in Pierre's second comment to get a list of GUIDs for all authors (having changed the limit to 10k) then iterated over them getting all of the properties from /people/person (D.O.B, nationality...), /book/author (list of books) and /common/type (image, article) in three different calls (oy). It works, though - again, as Pierre suggests. ;)

There is a way to get everything in one MQL call, though, as Alf says:

{
  "query" : [
    {
      "*" : null,
      "/book/author/books" : [
        {
          "*" : null
        }
      ],
      "/common/topic/article" : [
        {
          "*" : null
        }
      ],
      "/common/topic/image" : [
        {
          "*" : null
        }
      ],
      "limit" : 5,
      "profession" : "author",
      "type" : "/people/person"
    }
  ]
}

The disadvantage to this is that it's a lot of data to get in one go (considering there are thousands of authors each with lots of books)... I guess that's where paging through the results would come in (as Brendan correctly predicts).


And finally, for future reference.... skud points out that the mailing list is here.

Publishing house scale of web server evil

Kevin Burton recently checked to see which operating system the websites of different US presidential candidates are built on. The executive summary: Democrats use lots of Linux while Republicans (Ron Paul excepted) mainly use Windows.

Some have suggested a correlation between Windows web server usage and being evil*. This makes sense as only somebody with no soul could love ASP.

Does the theory hold true in the publishing world?

• PLoS run Linux
  
• Nature run Linux
  
• Science run Linux
  
• Wiley run Solaris
  
• Elsevier run Windows 2000
  
• Springer run Windows 2003
  

So from a purely progressive science on the web point of view.... yeah, sort of.

Springer, Elsevier and Wiley are pretty big companies and have lots of different sites, so maybe it's doing them a disservice to assume that whatever serves their root domain is their primary choice of OS. For example, I couldn't tell what Elsevier's ScienceDirect site runs because NetCraft returns 'unknown', so maybe it runs Linux.... or maybe NetCraft just doesn't have an entry for CRUSHED UP PUPPIES AND THE SWEAT OF THE OPPRESSED.

Trade publishing, for completeness:

• Canongate (arty Edinburgh based independent) run Linux
  
• Penguin (ironically) run Solaris 8
  
• Macmillan (publish Jeffrey Archer, employ me) run Windows 2000
  
• Simon & Schuster (publishing Lindsey Lohan's autobiography) run Windows 2003
  
• HarperCollins (owned by News International) run Windows 2003
  

* Don't read this the wrong way. Microsoft do lots of cool stuff nowadays too. But IIS is cold and heartless.

Nature on Facebook

It's in my Nature.com is Nature's splendid new Facebook group (there's a fan page, too, with a selection of fresh Nature.com content on it in case you need a quick fix of science news without leaving the 'book).

Anyway, it's splendid because it's a chance to interact with NPGers on an informal level - or rather a chance for us to interact with students & scientists on an informal level. The group started out as a mini-project run by a couple of Facebook early adopters and while it has picked up a lot of advice and support from within the company since then it's still a friendly place where everybody knows your name*.

Some of that company support has come in the form of free Nature-header-red iPods, so sign up and keep an eye out for giveaways in the near future.

(if you haven't seen it already you should also check out the PLoS group. After you've joined the Nature one, of course).

* though it's not the right place for ask the editor type questions

Where I lazily recycle news from OpenHelix

Geoff Bilder over at CrossRef has announced that their citation plugin for MT and WordPress is now available for download.

CrossRef is the shadowy cabal that runs the DOI system for journals. Shadowy because despite the fact that DOIs underpin academic publishing who outside of publishing tech circles has ever heard of them? Anyway, they've been doing a lot of cool science 2.0 stuff recently, probably because of Geoff, for whom I have a lot of respect. He does need to update his blog more often, though. ;)

(via OpenHelix)

Also at OpenHelix is a writeup of Gene Characterization Index: Assessing the Depth of Gene Annotation, published in PLoS One at the end of last year. It's a nice project.

One (very) minor niggle: the whole genome dataset they provide looks like this:

Gene ID GCI
2 10.0
19 10.0
24 10.0
25 10.0
...

WTF kind of gene ID?

It is blatantly obvious once you've read the actual paper (they're from Entrez), of course, but still. Dudes, some better column names or /* descriptive comments */ wouldn't go amiss.