A Generalized Method for Sentiment Analysis across Different Sources

Table 5

Results of Twitter sample dataset with average-IG feature and different classifiers.

Functions	TP	FP	TN	FN	Recall	Precision	F1	Accuracy
SIA	4658	1030	3970	342	93.2%	81.9%	87.2%	86.3%
ME	617	3805	1195	4383	12.3%	14.0%	13.1%	18.1%
NB	441	427	4573	4559	8.8%	50.8%	15.0%	50.1%
LR	4278	879	4121	722	85.6%	83.0%	84.2%	84.0%
MLP	4762	557	4343	338	93.4%	89.5%	91.4%	91.1%
Adaboost	3844	855	4145	1156	76.9%	81.8%	79.3%	79.9%
SVM	4674	571	4329	426	91.7%	89.1%	90.4%	90.0%

The bold values indicate the best performing entry along the column which is helpful for the reader.