Research Article
Enhancing Business Intelligence by Means of Suggestive Reviews
Table 4
Performance numbers on LR over all feature construction approaches for sentiment analysis of suggestive reviews.
| Feature construction | Precision | Recall | |
| Bow | 0.636 | 0.679 | 0.645 | Bow + Neg | 0.632 | 0.657 | 0.641 | Bow + Neg + Msk | 0.628 | 0.648 | 0.634 | Bow + Neg + Msk + Lex | 0.655 | 0.693 | 0.662 | Bow + Lex | 0.628 | 0.658 | 0.639 | Lex | 0.581 | 0.665 | 0.568 | Mix | 0.629 | 0.678 | 0.628 | Mix + Lex | 0.632 | 0.669 | 0.638 | Mix + Msk | 0.616 | 0.662 | 0.629 | Mix + Neg | 0.608 | 0.652 | 0.621 | Mix + Neg + Msk | 0.634 | 0.666 | 0.641 | Mix + Neg + Msk + Lex | 0.626 | 0.672 | 0.635 |
|
|