Old detectd information

detectd.html

       <table>
          <tr>
            <td>
              <div class="final">
                Discover events algorithmically detected right now!<br/>
                <span class="sub">
                  We currently detect 1000+ events everyday.
                  <a id="howitworks">How it Works</a>
                </span>
              </div>
            </td>

            <td>
              <div style="height: 40px; margin-top: 0px; width: 100px;">
                <div>
                  <a href="https://twitter.com/share" 
                     class="twitter-share-button" data-via="teleportd" data-related="teleportd">Tweet</a>
                  <script>!function(d,s,id){var js,fjs=d.getElementsByTagName(s)[0];if(!d.getElementById(id)){js=d.createElement(s);js.id=id;js.src="//platform.twitter.com/widgets.js";fjs.parentNode.insertBefore(js,fjs);}}(document,"script","twitter-wjs");</script>
                </div>
                <div style="">
                  <iframe frameborder="no" scrolling="no" height="50px" width="350px" src="http://hnlike.com/upvote.php?link=http%3A%2F%2Fdetectd.teleportd.com&title=Online+algorithmic+event+detection+from+realtime+mobile+photos">
                    iframes not supported by your browser
                  </iframe>
                </div>
              </div>
            </td>
          </tr>
        </table>

        <div class="howitworks">
          <br/><br/>

          <p>
            This event detector is based on the online analysis of the /stream endpoint of
            <a href="http://teleportd.com/api" target="_blank">Teleportd's API</a> (~ realtime aggregation of
            80+ mobile photos / seconds shared on public social networks. Think twitter,
            instagram, foursquare, etc...). Realtime geotagged pictures are a new and unique
            dataset to work on for event dectection. Indeed you can talk about anything in a
            geo-tagged tweet but you can only express an interest for things around you
            (<20m away) in a geo-tagged picture!
          </p>
          <p>
            To filter "interesting" events from this stream, we evolved the well-known DBSCAN algorithm
            (Density-Based Clustering) which is an OFFLINE algorithm into an ONLINE version which
            incrementally grows clusters as geo-tagged and timestamped data points arrive. Of course we
            also added the ability to work with the time dimension on top of the spacial ones. Interestingly,
            the attentive reader will probably realize that this algorithm is able to detect that a cluster
            won't be able to grow anymore, meaning the termination of an event for us! Which means we not only
            detect interesting events but also timestamp their beginning and their end!
          </p>
          <p>
            We were still having a lot of false-positives (we still have a few ones) so
            we built a compressed quadtree representation of the moving-average density of mobile
            photos worldwide to infer DBSCAN parameters instead of using constant ones (detection parameters
            can't be the same downtown NYC and in the middle of a desert!). There are a whole lot of other tricks
            that were used to build this demo, but we still have a long way to go and we'll be working hard
            until we go out of alpha!
          </p>
        </div>

    <script type="text/javascript">
      $(document).ready(function() {
        $('#howitworks').click(function() {
          console.log('YEAH');
          $('.howitworks').toggleClass('visible');
        });
      });
    </script>