Merging SenticNet and WordNet-Affect emotion lists for sentiment analysis
94 Citations•2012•
Soujanya Poria, Alexander Gelbukh, E. Cambria
2012 IEEE 11th International Conference on Signal Processing
This work reports a work on automatically merging SenticNet and WordNet-Affect by assigning emotion labels to more than 2700 concepts, which is a major step towards merging these two resources.
Abstract
SenticNet is currently one of the most comprehensive freely available semantic resources for opinion mining. However, it only provides numerical polarity scores, while more detailed sentiment-related information for its concepts is often desirable. Another important resource for opinion mining and sentiment analysis is WordNet-Affect, which in turn lacks quantitative information. We report a work on automatically merging these two resources by assigning emotion labels to more than 2700 concepts.