Code Camp Now Suggests Sessions you MIGHT want to Attend!

All Code Camp Sessions have multiple tags (like JSON,Java,C#,XBox,Gaming, etc.).  These tags have been used for many years on many many sessions.  You can imagine that when you indicate you plan to attend a session (or even that you are interested), you are tacitly telling code camp you are possibly interested in any sessions that might have that same tag in them.  Well, we now surface that for you in the Sessions Overview page.  If you are logged in and have indicated you plan to attend 1 or more sessions, almost guaranteed, you will now see in purple some suggestions. 

 

image

 

 

Just to Illustrate this point, let’s say you are a first time code camp user.  You have just created your account and as an avid game developer you find Bill Crow has a session titled Create a Kinect Powered Personal Robot with Microsoft Robotics Developer Studio  You indicate that you plan to attend it by checking that box on the Sessions page.

 

image

 

Notice that you have indicated you want to go to a session that has 7 different tags.  C#, Kinect, NET, etc.  Now, when you go to the Sessions Overview page, you will see that you have indicated you plan on attending Bills’s session by the words (Plan On Attending) next to his  session, and also notice that tens of other sessions have been highlighted including “Introduction to the XBox Kinect SDK from Microsoft”  This is because both of these sessions were tagged with Kinect.

 

image

 

image

 

Currently, lots of similar sessions are suggested, and since we have lots of general tags like (Mobile), quickly, almost all sessions are suggested.  We are working on that and if you’ve got any ideas on how to restrict the choice to make better suggestions, please let us know by commenting below.  Maybe together, we can make this really useful.

Tags: , ,

2 People have left comments on this post



» AlexandreZani said: { Sep 27, 2011 - 08:09:51 }

Why not compute a simple relevance score for each session keyword? Compute 1 minus the percentage of the sessions that sport the keyword. Then, compute the percentage of the sessions containing that keyword which the attendee is signed up for. Multiply the two numbers to find the score for that attendee-keyword pair. Then add up all the attendee-keyword pairs for a particular session to calculate that session score.

E.g:

90% of sessions have the keyword Mobile. 10% have the keyword Python. An attendee is attending 50% of the Python sessions and 10% of the Mobile sessions. A session with both Python and Mobile would be scored as (1-0.9) * 0.1 + (1-0.1) * 0.5 = 0.46. Normalize all the sessions for an attendee and pick the top 10% best scoring or something like that.

It’s far from perfect, subject to gaming etc, but it might be a step in the right direction…

» AlexandreZani said: { Sep 27, 2011 - 08:09:53 }

Actually it might be more useful to use the percentage of the sessions attended by an attendee that contain a keyword. So for instance, we might look at the fact that 20% of the sessions an attendee signed up for contain the keyword Python rather than looking at the fact that an attendee is attending 20% of the sessions that contain the Python keyword.