EmoText: Applying Differentiated Semantic Analysis in Lexical Affect Sensing

1
EmoText Applying Differentiated Semantic
Analysis in Lexical Affect Sensing

• Alexander Osherenko
• University of Augsburg

2
Examples

• I am very happy
• I am not happy
• I am very happy and sad
• I am very sad even if everybody is happy!
• I am very sad, but everybody else is happy!

3
EmoText

• Deployment as Webinterface/Webservice
  (http//emotext.informatik.uni-augsburg.de)
• Relies on deep parsing in the Stanford parser and
  the SPIN parser
• Emotion words from WordNet-Affect (DAL), GI,
  Levin verbs
• Linguistic intensifiers and negations, empirical
  grammer rules with superordinate/subordinate
  clauses

Comprehensive description of the approach
tomorrow, 11.09.2009 at 11.25