cho3 / AffectiveClassification

Affective Classification

Experiments in regression and classification on text-based affect (emotion) datasets.

Benchmarks

Another person did experiments on both the ISEAR and Fairtale (Alm2007) datasets as well, https://github.com/bogdanneacsa/tts-master TODO: list semeval2007

Valence

Method	r	Acc	Prec	Rec	F1
semeval-best	47.70	55.10	61.42	66.38	42.43

Anger

Method	r	Acc	Prec	Rec	F1
semeval-best	32,33	93.60	16.67	21.60	16.03

Disgust

Method	r	Acc	Prec	Rec	F1
semeval-best	18.55	97.30	0	0	N/A

Fear

Method	r	Acc	Prec	Rec	F1
semeval-best	44.92	87.90	33.33	26.27	20.06

Joy

Method	r	Acc	Prec	Rec	F1
semeval-best	26.11	82.20	54.54	9.44	14.91

Sadness

Method	r	Acc	Prec	Rec	F1
semeval-best	40.98	89.00	48.97	22.92	30.38

Surprise

Method	r	Acc	Prec	Rec	F1
semeval-best	16.71	89.10	13.70	16.56	15.00

Trust

Method	r	Acc	Prec	Rec	F1
semeval-best	N/A	N/A	N/A	N/A	N/A

Anticipation

Method	r	Acc	Prec	Rec	F1
semeval-best	N/A	N/A	N/A	N/A	N/A

Datasets

TODO: list datasets