Table 9 Results of traditional machine learning experiments with various n-grams in the English language.
From: Multilingual identification of nuanced dimensions of hope speech in social media texts
Models | LR | SVM(Linear) | SVM(RBF) | MNB | DT | RF |
|---|---|---|---|---|---|---|
Binary Hope Speech Detection | ||||||
Unigrams | 0.7957 | 0.7920 | 0.8006 | 0.7322 | 0.7372 | 0.7959 |
Bigrams | 0.6577 | 0.7026 | 0.6367 | 0.7015 | 0.6124 | 0.6530 |
Trigrams | 0.5444 | 0.5644 | 0.5183 | 0.5750 | 0.5495 | 0.5384 |
Uni+Bi-grams | 0.7943 | 0.7997 | 0.7983 | 0.7649 | 0.7500 | 0.7873 |
Bi+Tri-grams | 0.6334 | 0.6834 | 0.5986 | 0.6834 | 0.6128 | 0.6310 |
Uni+Bi+Tri-grams | 0.7911 | 0.7984 | 0.7938 | 0.7624 | 0.7528 | 0.7786 |
Multiclass Hope Speech Detection | ||||||
Unigrams | 0.4863 | 0.5169 | 0.4496 | 0.2507 | 0.4661 | 0.4649 |
Bigrams | 0.2555 | 0.3551 | 0.2015 | 0.2175 | 0.3980 | 0.3590 |
Trigrams | 0.1771 | 0.1960 | 0.1739 | 0.1748 | 0.2587 | 0.2169 |
Uni+Bi-grams | 0.4203 | 0.4983 | 0.3729 | 0.2141 | 0.4751 | 0.4310 |
Bi+Tri-grams | 0.2074 | 0.2700 | 0.1791 | 0.1906 | 0.3971 | 0.3359 |
Uni+Bi+Tri-grams | 0.3686 | 0.4613 | 0.3150 | 0.1954 | 0.4741 | 0.4134 |
