A hybrid feature selection method is proposed to distinguish the salient features that allow identifying the viewpoint underlying a text review, that is, to determine its sentiment polarity. This method makes use of fundamental pre-processing tasks known as filter and wrapper techniques. The effectiveness of this approach is demonstrated on a data set where each document is represented by two distinct feature vectors based on two different sets of rules.
Optimizing Feature Selection Techniques for Sentiment Classification
2011-11-01
149430 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Texture classification based on evolution feature selection [4554-33]
British Library Conference Proceedings | 2001
|