A Performance Comparison of Unsupervised Techniques for Event Detection from Oscar Tweets

Table 4

Topic recall, keyword precision, and keyword recall of all the methods.

Methods	Number of topics	Topic recall	Keyword precision	Keyword recall
LDA	5	0.6	1	0.6
K-SVD	5	0.8	1	1
Doc-p	5	1	1	1
Feat-p	5	0.6	0.8	0.6
SFPM	5	1	1	1
LDA	10	0.8	0.7	0.7
K-SVD	10	0.8	0.8	0.8
Doc-p	10	0.8	1	0.8
Feat-p	10	0.7	0.7	0.6
SFPM	10	0.8	0.9	0.8
LDA	15	0.5	0.7	0.5
K-SVD	15	0.7	0.8	0.7
Doc-p	15	0.5	0.9	0.5
Feat-p	15	0.5	0.7	0.5
SFPM	15	0.8	0.9	0.8
LDA	20	0.6	0.6	0.5
K-SVD	20	0.7	0.8	0.6
Doc-p	20	0.6	0.8	0.5
Feat-p	20	0.6	0.6	0.5
SFPM	20	0.8	0.9	0.8
LDA	25	0.5	0.5	0.4
K-SVD	25	0.6	0.7	0.5
Doc-p	25	0.5	0.5	0.5
Feat-p	25	0.5	0.5	0.4
SFPM	25	0.7	0.7	0.6