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

an algorithm for discovering blogversations

Expand Messages
  • Lucas Gonze
    This algorithm isn t quite right, or maybe it is except that I m not putting the energy into implementing it. The goal is to discover blogversations. You want
    Message 1 of 4 , May 13, 2004
    • 0 Attachment
      This algorithm isn't quite right, or maybe it is except that I'm not
      putting the energy into implementing it.

      The goal is to discover blogversations. You want to find all the entries
      on different blogs related to the same topic. Ideally you can also
      provide a semi-threaded view of them.

      Assume a service like TopicExchange and some but not absolute self
      awareness among participants in the conversation. The service allows web
      authors to create URLs which uniquely identify topics. There is one URL
      per conversation. These aren't huge categories like "politics", they're
      small ones like "Bob's idea about charsets."

      When I blog a topic, I include a link from my entry to the TopicExchange
      entry, so that when somebody clicks through TopicExchange can log the
      referrer as a participant in the conversation. Then, combine the referrer
      logs with a shallow spidering (by TopicExchange) of referrers to discover
      malingerers who don't link to the topic. Lastly, when the spider discovers
      a nearby neighbor who has also linked to the topic URL, consider that
      neighbor strongly connected. You'll end up with three classes of nodes:
      strongly connected ones who link to the topic, weakly connected ones who
      are linked from and to strongly connected ones, and fully unconnected ones
      like search engines.

      This would do two good things. It would distinguish bots from human
      agents, and it would bring in participants who don't link to the topic.

      ?

      - Lucas
    • Francis Hwang
      Lucas, I found this intriguing but I got confused by a few of the details. ... Which referrer logs are you analyzing? TopicExchange s? The original poster s?
      Message 2 of 4 , May 19, 2004
      • 0 Attachment
        Lucas,

        I found this intriguing but I got confused by a few of the details.
        Well, a lot of the details. Here are some questions:

        > Assume a service like TopicExchange and some but not absolute self
        > awareness among participants in the conversation. The service allows
        > web
        > authors to create URLs which uniquely identify topics. There is one
        > URL
        > per conversation. These aren't huge categories like "politics",
        > they're
        > small ones like "Bob's idea about charsets."
        >
        > When I blog a topic, I include a link from my entry to the
        > TopicExchange
        > entry, so that when somebody clicks through TopicExchange can log the
        > referrer as a participant in the conversation. Then, combine the
        > referrer
        > logs with a shallow spidering (by TopicExchange) of referrers to
        > discover
        > malingerers who don't link to the topic. Lastly, when the spider
        > discovers
        > a nearby neighbor who has also linked to the topic URL, consider that
        > neighbor strongly connected. You'll end up with three classes of nodes:
        > strongly connected ones who link to the topic, weakly connected ones
        > who
        > are linked from and to strongly connected ones, and fully unconnected
        > ones
        > like search engines.

        Which referrer logs are you analyzing? TopicExchange's? The original
        poster's? Both?

        What do you mean by "malingerers who don't link to the topic"? Do you
        mean "bloggers who are participating in the conversation but not
        linking to TopicExchange"?

        What specific needs would we expect to fulfull that can't be fulfilled
        with things like Trackbacks?

        Francis Hwang
        http://fhwang.net/
        AIM: francisrhizome
      • Lucas Gonze
        Hey Francis, ... TopicExchange s. ... Yes, exactly. ... One difference between this method and a trackback is that the connection is recorded just by clicking
        Message 3 of 4 , May 19, 2004
        • 0 Attachment
          Hey Francis,

          On Wed, 19 May 2004, Francis Hwang wrote:
          > Which referrer logs are you analyzing? TopicExchange's? The original
          > poster's? Both?

          TopicExchange's.

          > What do you mean by "malingerers who don't link to the topic"? Do you
          > mean "bloggers who are participating in the conversation but not
          > linking to TopicExchange"?

          Yes, exactly.

          > What specific needs would we expect to fulfull that can't be fulfilled
          > with things like Trackbacks?

          One difference between this method and a trackback is that the connection
          is recorded just by clicking through. There's no need to initiate a cgi
          type of interaction.

          Another difference is that, on a dreamy not quite right level, I'm hoping
          to scoop up the malingerers as well as people who get directly involved.
          It should be possible to map the whole conversation without having every
          participant formally announce that they're involved.

          - Lucas
        • Francis Hwang
          ... This is close to one of the problems I have with all the various ways of aggregating and searching online conversations: They re really blog-centric. When
          Message 4 of 4 , May 24, 2004
          • 0 Attachment
            Lucas wrote:

            > Another difference is that, on a dreamy not quite right level, I'm
            > hoping
            > to scoop up the malingerers as well as people who get directly
            > involved.
            > It should be possible to map the whole conversation without having
            > every
            > participant formally announce that they're involved.

            This is close to one of the problems I have with all the various ways
            of aggregating and searching online conversations: They're really
            blog-centric. When I look at referer logs for my own site I see lots of
            interesting content that couldn't really be considered a blog: They've
            got no RSS, maybe you're dealing with a monthly zine with big articles
            instead of twenty snippets a day. A few of my articles on my site are
            getting used in class curricula, which somehow seems as interesting a
            point of conversation as that of a blogger's referring to your page.

            The problem with defining new protocols, though, is that implementation
            of anything non-trivial is expensive (compared to the very low cost, in
            terms of time, that it takes to write online content of any sort).
            Perhaps even collecting a TopicExchange-type URL might be a lot of
            work. It's also just not so much fun, which is always a problem.
            Personally I'm starting to think that there's a lot more to be gleaned
            out of what we've already got.

            One big item on my techie to-do list is to analyze referer logs to
            figure out who's posting references to various pages on my site. I
            could imagine a tool that I use once a week or so, to approve or delete
            for spam. (Personally I'm not really hot on the 10-posts-a-day blogging
            model: I don't have the time.) It would be basically like Trackbacks,
            only it would suck in everything else, too: wikis, class curricula,
            online magazines, etc. Heck, anything that can be done to shake up the
            bloggers echo chamber is healthy.

            Francis Hwang
            http://fhwang.net/
            AIM: francisrhizome
          Your message has been successfully submitted and would be delivered to recipients shortly.