Table 11 Classification results (macro averaged F1-score) using dictionary features. Results on WH+TW are averaged on ten samples
From: Analysis and classification of privacy-sensitive content in social media posts
Dataset | Class. | All dict. | Psych. dict. | Priv. Dict |
|---|---|---|---|---|
SENS2 | LR | 0.64 | 0.65 | 0.38 |
RF | 0.65 | 0.66 | 0.41 | |
SVM | 0.64 | 0.65 | 0.38 | |
SENS3 | LR | 0.72 | 0.72 | 0.39 |
RF | 0.70 | 0.69 | 0.42 | |
SVM | 0.70 | 0.72 | 0.39 | |
OMC | LR | 0.63 | 0.63 | 0.38 |
RF | 0.67 | 0.66 | 0.40 | |
SVM | 0.62 | 0.63 | 0.38 | |
WH+TW | LR | 0.78 ± 0.01 | 0.77 ± 0.01 | 0.46 ± 0.01 |
RF | 0.78 ± 0.01 | 0.78 ± 0.01 | 0.49 ± 0.02 | |
SVM | 0.77 ± 0.01 | 0.77 ± 0.01 | 0.46 ± 0.01 |