Table 18 Cross-classification results (macro-averaged F1-scores). Classifiers are trained on the datasets reported in the row, and tested on the datasets reported in the columns
From: Analysis and classification of privacy-sensitive content in social media posts
Dataset | Class. | SENS2 | SENS3 | OMC | WH+TW |
|---|---|---|---|---|---|
SENS2 | DICT-RF | – | 0.66 | 0.38 | 0.46 ± 0.00 |
BoW | – | 0.75 | 0.44 | 0.50 ± 0.00 | |
BERT | – | 0.90 | 0.50 | 0.58 ± 0.00 | |
SENS3 | DICT-RF | 0.63 | – | 0.37 | 0.44 ± 0.00 |
BoW | 0.64 | – | 0.42 | 0.47 ± 0.00 | |
BERT | 0.74 | – | 0.48 | 0.51 ± 0.00 | |
OMC | DICT-RF | 0.33 | 0.21 | – | 0.52 ± 0.01 |
BoW | 0.51 | 0.51 | – | 0.55 ± 0.00 | |
BERT | 0.56 | 0.59 | – | 0.52 ± 0.01 | |
WH+TW | DICT-RF | 0.29 ± 0.00 | 0.18 ± 0.00 | 0.54 ± 0.01 | – |
BoW | 0.48 ± 0.01 | 0.47 ± 0.02 | 0.50 ± 0.01 | – | |
BERT | 0.45 ± 0.01 | 0.44 ± 0.01 | 0.57 ± 0.01 | – |