Affective Usenet

The typical usenet experience boils down to sifting through massive volumes of text.   In these communal sub-nets there are so many anonymous trolls, unending semantic disputes and arguments that each click is a potential surprise. Affective Usenet is a tool to figure out the affective quality of a usenet message along side other header information.

The project uses semantic data from ConceptNet to sift through messages and estimate the emotional valence of the overall message. The program replaces a standard field from Usenet messages (X-Client) with the emotional value of the document. It uses one of the 8 different values on Albert Mehrabian's PAD (Pleasure/Arousal/Dominance) affective scale, and SHOUTS THEM (uppercase) when they are intense.

NOTE: This only changes the content of source newsgroup data. Many usenet clients only download the headers, so make sure you have one that is suitable for 'offline' use -- and most keep messages separate from an index file that has some of the header information. On my OSX Mac, I used Pineapple News because of its supported downloading message bodies and re-indexing on command.

Download Affective Usenet. (12Mb)



Once you download, run the mailread.rb file on newsgroup files: 
%> ruby mailread.rb file1 file2 ...
It will overwrite the input files with files of the same name and content, except the change in the X-Client header value.

For Pineapple users: 
%> ruby mailread.rb ~/Library/PineappleData/News/Accounts/*/Groups/*/*.pmsg


Thanks to:

Ben Bongalon and his RubyCon.
Hugo Liu and his ConceptNet.
The emotionally charged people on comp.ai.philosophy 		≈
Affective Usenet in Action!

Usenet or Abusenet?
(click for closeup)

Back to Dustin's homepage.