Modeling social media conversations?

[Norman Walsh <ndw@nwalsh.com>]

Hello world,

Has anyone done any work modeling social media conversations (twitter,
Google+, Facebook, etc.)?

For my "short form" essay collections, I've been modeling Twitter
conversations as nested lists with role attributes. Modeling Google+
postings in the same way would be much harder and would require a lot
more brute force.

I'm somewhat tempted to step back and build a common model for both.
There isn't perfect alignment, but it's not too hard to see a Twitter
status message as a kind of very simple posting where Google+ allows
richer postings.

For example:

<posting>
  <service>googleplus</service>
  <id>z135gjeh1rmuw1tt423jsze54mbrj10pp04</id>
  <verb>post</verb>
  <type>note</type>
  <pubdate>2012-01-23T12:26:55.000Z</pubdate>
  <link>https://plus.google.com/112652695581272951676/posts/7A6CoUwZqN9</link>
  <title>Apologies for the noise. I'll try not to do this too often.
  Just testing my Google+ archiving scr...</title>
  <user>
    <id>112652695581272951676</id>
    <name>Norman Walsh</name>
    <link>https://plus.google.com/112652695581272951676</link>
    <avatar>https://lh3.googleusercontent.com/-dw7Amoq9XCA/AAAAAAAAAAI/AAAAAAAAAAA/4Tpu-uUaB14/photo.jpg?sz=50</avatar>
  </user>
  <content>Apologies for the noise. I'll try not to do this too often.
  Just testing my Google+ archiving script.</content>
  <attachments>
    <attachment>
      <type>article</type>
      <link>http://norman.walsh.name/2012/01/22/shortform</link>
      <name>Google+ Postings</name>
      <content>Nice epitaph: Beyond the universe there is nothing and
within the universe the supernatural does not and cannot exist. Of all
deceivers who have plagued mankind, none are so deeply ruinous to
human ha...</content>
    </attachment>
    <attachment>
      <type>photo</type>
      <link>http://images0-focus-opensocial.googleusercontent.com/gadgets/proxy?container=focus&gadget=a&resize_h=100&url=http%3A%2F%2Fchart.apis.google.com%2Fchart%3Fchs%3D125x125%26cht%3Dqr%26chl%3Dhttp%3A%2F%2Fnorman.walsh.name%2F2012%2F01%2F22%2Fshortform</link>
    </attachment>
  </attachments>
</posting>

Or

<posting>
  <service>twitter</service>
  <id>161549754021584896</id>
  <verb>posting</verb>
  <pubdate>2012-01-23T20:43:57+00:00</pubdate>
  <link>https://twitter.com/xmlprague/status/161549754021584896</link>
  <user>
    <id>15942706</id>
    <name>xmlprague</name>
  </user>
  <content>please send us your questions for the 'Panel discussion on HTML/XML convergence' #<a class="tag" href="http://twitter.com/search?q=%23xmlprague";>xmlprague</a> #<a class="tag" href="http://twitter.com/search?q=%23html";>html</a> #<a class="tag" href="http://twitter.com/search?q=%23xml";>xml</a>
  </content>
  <reposted>true</reposted>
</posting>

Not sure how to capture the conversational nature of Twitter status
messages. And there are some semantic gaps in the data returned by the
relevant APIs, but it doesn't have to be perfect.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com>      | The fundamental delusion of
http://www.oasis-open.org/docbook/ | humanity is to suppose that I am
Chair, DocBook Technical Committee | here and you are out
                                   | there.--Yasutani Roshi

Attachment: pgp0WA9onp1ht.pgp
Description: PGP signature