Loading ...
Sorry, an error occurred while loading the content.

Re: [govtrack] GovTrack Machine Tags

Expand Messages
  • Josh Tauberer
    ... Hi, Joe. I m not a particularly big fan of tags. I see them as a very short term, unorganized solution to a much larger problem. As time goes on, the
    Message 1 of 5 , Jun 18, 2007
    • 0 Attachment
      Joe Germuska wrote:
      > Has there been any discussion of a machine tag convention for GovTrack
      > "nodes"? (see http://machinetags.org/ or
      > http://en.wikipedia.org/wiki/Machine_tag
      > <http://en.wikipedia.org/wiki/Machine_tag>) If not, is there any interest?

      Hi, Joe.

      I'm not a particularly big fan of tags. I see them as a very short term,
      unorganized solution to a much larger problem. As time goes on, the
      likelihood of cashes of tag names grows and grows (i.e. ambiguity of
      what the tag was meant to refer to), and as people try to push more
      information inside of a tag, we get stuck trying to parse the
      information back out, not knowing whether it was intended or not.

      My preferred approach is using semantic web methods, like URIs for
      identifying things and RDF for encoding structured data.

      > I could imagine a lot of value to the community of GovTrack users
      > tagging del.icio.us <http://del.icio.us> bookmarks or whatnot with tags
      > that could be linked back to specific bills.

      And so from that perspective, I'm not opposed to adding things like that
      to the site, but people would have to point me in the right direction
      for what to do.

      > govtrack:bill=h110-2206

      In the ideal world, I would use a globally unique identifier for each bill.

      > One could also suggest a format for "people" tags, like
      > govtrack:person=300038 (Dick Durbin) Not sure if there is any usable
      > unique ID that would make it more "shared" than govtrack.

      There are IDs assigned with the Library of Congress (see
      bioguide.congress.gov) which are neutral ground, although they have some
      issues like multiple identifiers assigned to the same individual ---
      after women change their name (at least in their historical data, which
      might just be unintentional).

      > For people like Dick Durbin, the likelihood of users selecting the same
      > tag intuitively is much higher than for tagging a bill.

      And the likelihood that other Dick Durbin's in the world get conflated
      with our congressional guy also is higher, which is why I don't like
      tags much.

      But like I said, I'm not deeply opposed to using them on the website.

      - Josh
    • Joe Germuska
      ... Josh: Thanks for your response. I understand your preference for unambiguous semantic annotations. I am of mixed mind myself, as I feel that the populist
      Message 2 of 5 , Jun 18, 2007
      • 0 Attachment
        On 6/18/07, Josh Tauberer <tauberer@...> wrote:

        Joe Germuska wrote:
        > Has there been any discussion of a machine tag convention for GovTrack
        > "nodes"? (see http://machinetags.org/ or
        > http://en.wikipedia.org/wiki/Machine_tag
        > <http://en.wikipedia.org/wiki/Machine_tag>) If not, is there any interest?

        Hi, Joe.

        I'm not a particularly big fan of tags. I see them as a very short term,
        unorganized solution to a much larger problem.

        Josh:

        Thanks for your response.

        I understand your preference for unambiguous semantic annotations.  I am of mixed mind myself, as I feel that the populist potential of tags is important.  The problem is whether machine tags are simple enough that they'll be popularly adopted, or whether they will end up not gaining traction. 

        Or, the problem (if one is not particularly populist) may be that users can't be trusted to tag things correctly, even if tags are not considered so ambiguous.  However, if taggers adhere to a contract, there is no reason that any given tag can't be mapped to a URI appropriate for use in an RDF triple subject or object.

        So, one could easily say that, within the machine tag convention,
        govtrack:person=300038
        means exactly
        tag:govshare.info,2005:data/us/congress/people/D000563

        so that you could read a feed of Delicious bookmarks and construct RDF triples from them that say
        (A New York Times article) (is about) (Dick Durbin).  Is "aboutness" interesting enough?  I have trouble believing that even experts are going to make much more meaningful relationships.  [ (A Drudge Report posting) (slanders) (a member of congress) (A letter to the editor) (totally misrepresents) (the bill to outlaw television) ?? ]  But even if they were going to, one could come up with a taggish compression pairing common predicates with well-identified objects. 

        I know you don't care for the term "tagging", but I can't think of anything else to call "every day people annotating pages they come across."  Do you see any general use cases for human-mediated structuring of (gardening?) the pool of data that makes up GovTrack?  Right now, everything, or just about, is generated by transforming data files, correct?  Are there places where human intelligence would make something better? 

        If there are, then there need to be tools to help -- no one writes RDF with a text editor for very long.  But there's no reason one couldn't use tag autocompletion and human-readable names to make it pretty easy for people to contribute data "from the field," if there's data that is to be desired.

        In one sense, this isn't much different from using the Technorati feed of people who have linked to the bill.  But realistically, maybe just the effort devoted to writing a post about a bill vs. tag-and-run ends up being a useful way to keep the signal-to-noise ratio better.

        but now I'm starting to ramble...

        Joe

        (PS Is GovTrack data part of Freebase.com's data pile?)


        --
        Joe Germuska
        Joe@... * http://blog.germuska.com   
        "I felt so good I told the leader how to follow."
        -- Sly Stone


      • Bryan L. Fordham
        Personally, I love the idea of tagging, though not necessarily as something that would be on the front page. It s on my todo list at critter watch, though
        Message 3 of 5 , Jun 19, 2007
        • 0 Attachment
          Personally, I love the idea of tagging, though not necessarily as
          something that would be on the front page. It's on my todo list at
          critter watch, though admittedly it's pretty far down the list.

          But yeah, there are a number of problems with it. And some good things.
          I think letting folks tag items, and other users can optionally view
          them (or perhaps view the most popular tags for some items) would be a
          fun experiment.

          --B
        • Joe Germuska
          Just to be clear ... Just for clarity s sake, let me point out that I m not talking at all about GovTrack allowing users to apply tags to GovTrack data. I
          Message 4 of 5 , Jun 19, 2007
          • 0 Attachment
            Just to be clear

            On 6/19/07, Bryan L. Fordham <bfordham@...> wrote:

            Personally, I love the idea of tagging, though not necessarily as
            something that would be on the front page. It's on my todo list at
            critter watch, though admittedly it's pretty far down the list.

            But yeah, there are a number of problems with it. And some good things.
            I think letting folks tag items, and other users can optionally view
            them (or perhaps view the most popular tags for some items) would be a
            fun experiment.



            Just for clarity's sake, let me point out that I'm not talking at all about GovTrack allowing users to apply tags to GovTrack data.  I don't have a very strong opinion about whether or not GovTrack should support that. 

            Take, for example S. 1348: A bill to provide for comprehensive immigration reform and for other purposes GovTrack already shows us ten blogs that have linked to the tracking page for that bill, using Technorati. 

            But not all sources will know to (or choose to) link to the GovTrack page.  What if I'm passionate about this issue and I find this article on MSNBC about the bill.  If I knew how to tag it reliably in Delicious, then Josh could set up another feed alongside the Technorati one which would pick up on my recognition that MSNBC was talking about S. 1348, even though MSNBC didn't link back to GovTrack, or even use the bill number in the article.  Of course, with the right convention, this is not limited to GovTrack by any means.  Political blogs of any persuasion could do the same thing.

            Now, maybe one of Josh's hesitations is the question about how many taggers in the field would make the correct links, and that's a valid question.  For blogs with more of an editorial position, I think the simple answer is to limit the number of users whose tags are even considered.  And maybe at the end of the day, GovTrack itself wouldn't want to make those choices and wouldn't want to accept a totally unfiltered tag feed (although the Delicious spammers would probably use a lot of other tags first before they start abusing wonky machine-oriented tags).  But since this is a list of people who are generally interested in applying technology to civics, I thought I'd float it...

            I hope the above example clarifies my point.  Should anyone be at all interested, I wrote a few things about this general topic that have drifted off the front page of my blog:
            http://blog.germuska.com/?p=496
            http://blog.germuska.com/?p=497

            Joe

            --
            Joe Germuska
            Joe@... * http://blog.germuska.com    

            "I felt so good I told the leader how to follow."
            -- Sly Stone
          Your message has been successfully submitted and would be delivered to recipients shortly.