Table 7 Performance rank of TPF based methods in six datasets
Classifier | S. No. | Dataset | Maximum accuracy achieved (%) | Number of features | Method |
|---|---|---|---|---|---|
MNB | 1. | movie review | 98.4 | 10,000 | \(\hbox {SSNG} + \chi ^2\) |
2. | ACL IMDB | 89.81 | 20,000 | \(\hbox {SSNG} + \chi ^2\) | |
3. | Ohsumed5 | 84.03 | 1000 | \(\hbox {SSNG} + \chi ^2\) | |
4. | Ohsumed10 | 67.32 | 2000 | \(\hbox {SSNG} + \chi ^2\) | |
5. | Ohsumed15 | 43.91 | 2000 | \(\hbox {SSNG} + \chi ^2\) | |
6. | Ohsumed23 | 43.91 | 2000 | \(\hbox {SSNG} + \chi ^2\) | |
7. | Pubmed9 | 73.84 | 5000 | \(\hbox {SSNG} + \chi ^2\) | |
8. | 20Newsgroup | 95.6 | 500 | \(\chi ^2+ \chi ^2\) | |
9. | Reuters13 | 71.59 | 500 | \(\chi ^2+ \chi ^2\) | |
10. | BBC_Sports | 98.39 | 500, 1000, and 2000 | \(\hbox {SSNG} + \chi ^2\) | |
11. | BBC | 99.28 | 1000, 5000 | \(\hbox {IG} + \chi ^2\), \(\hbox {SSNG} + \chi ^2\) | |
LSVM | 1. | movie review | 95.8 | 3000, and 5000 | \(\hbox {SSNG} + \chi ^2\), and \(\hbox {SSNG} + \chi ^2\), \(\hbox {OR} + \chi ^2\) |
2. | ACL IMDB | 89.94 | 15,000 | \(\hbox {SSNG} + \chi ^2\) | |
3. | Ohsumed5 | 86.24 | 3000,10,000 | \(\hbox {SSNG} + \chi ^2\) | |
4. | Ohsumed10 | 70.18 | 15,000 | \(\hbox {SSNG} + \chi ^2\) | |
5. | Ohsumed15 | 65.75 | 10,000 | \(\hbox {SSNG} + \chi ^2\) | |
6. | Ohsumed23 | 48 | 15,000 | \(\hbox {SSNG} + \chi ^2\) | |
7. | Pubmed9 | 74.15 | 2000 | \(\hbox {SSNG} + \chi ^2\) | |
8. | 20Newsgroup | 95.8 | 3000, and 5000 | \(\hbox {SSNG} + \chi ^2\) | |
9. | Reuters13 | 78.52 | 2000 | \(\hbox {SSNG} + \chi ^2\) | |
10. | BBC_Sports | 100 | 500, 1000, and 3000 | \(\chi ^2+ \chi ^2\), \(\hbox {IG} + \chi ^2\), and \(\hbox {SSNG} + \chi ^2\) | |
11. | BBC | 99.64 | 10,000 | \(\hbox {SSNG} + \chi ^2\) |