Research Article
Feature Reduction Based on Genetic Algorithm and Hybrid Model for Opinion Mining
Table 1
Description of multidomain dataset and movie review dataset (unigram).
|
Dataset |
Number of stratified samples |
Positive reviews |
Negative reviews |
Total attributes | Total attributes (weight by IG) |
Total attributes (optimized selection) | NB, LR, SVM | BSVM | BNB |
| Book | 191 | 100 | 91 | 377 | 341 | 264 | 231 | 231 | DVD | 199 | 99 | 100 | 364 | 318 | 255 | 228 | 219 | Electronics | 200 | 100 | 100 | 274 | 209 | 192 | 169 | 175 | Kitchen | 190 | 99 | 91 | 207 | 193 | 145 | 123 | 131 | Movie | 200 | 100 | 100 | 1569 | 1568 | 1098 | 1069 | 958 |
|
|