Gephi has improved a lot

Gephi, an open source visualization software has improved a lot lately with their 0.7 beta release. I did not use this before because it could only create graphs and not analyze them, but now thing are looking quite good.

It appears to even have tools to merge datasets. It also has ability for time slicing.  :-D I will explore this further later on.

What we need now is for Moodle forums to be able to export node-tie data in GEXF format (Gephi's native file format). They also seem to have a flash viewer that can be put in a Moodle plugin, server side called GEXF Explorer. If this is correct all the pieces needed to come up with a periodic graph visualization tool for Moodle forums is there.  I think the best solution is simply to access Moodle's forum database for user id, post and reply data and come up with a page that allows server side visualization and export to GEXF for offline analysis. I don't think analysis should be done server side as it would slow down Moodle. It could be done if there is a dedicated server doing the number crunching.

By taking advantage of Gephi's dynamic data attributes the user can select a time period for analyzing data in a dropdown box, let's say daily, weekly, or monthly.

An additional improvement on tie-node data gathering for the Moodle forum is to have a filter that will parse nicknames like @John for John Doe. When a user types @John the editor will display somewhere (e.g. a block maybe) an offer to associate @John with another user e.g. John Doe, John Deer etc. Then the nickname @John or an image icon can appear in a cloud block so that the poster can simply click on that to insert a nickname. The filter will then insert a node-tie tag that identifies @John with John Doe. Example is when Jane Little replied to Peter Potamus' post but added a reply to John's other post in the post body. Then the filter can tag the reply to John as @John<displaytext="@John" userid="123" username="John Doe">. The from data can be taken from the reply user id.  This way we do not have to do content analysis to find lost ties.

The question now is who's willing and able to code this Moodle third party plugin?