Research Article
Fuzzy Aspect Based Opinion Classification System for Mining Tourist Reviews
Table 2
Critical evaluation of aspects based classification methods.
| Reference | Dataset (hotels, restaurants) | Two-point scale | Five-point scale | Method | Type | Prediction | Results |
| Colhon et al., 2014 [21] | Review: 2521 | Yes | | Opinion terms counting method | Lexicon based | Compared with user reviews results | 87% | Marrese-Taylor et al., 2014 [19] | Reviews: 200 | Yes | | Terms score method | Lexicon based | Compared with tourist experts results | 90% | Pekar and Ou, 2008 [15] | Reviews: 268 | | Yes | Opinion terms intensity method | Lexicon based | Compared with judges results | 78% | Marrese-Taylor et al., 2013 [18] | Reviews: 1435 | Yes | | Terms score method | Lexicon based | Compared with tourist experts results | 83% | Muangon et al., 2014 [16] | Reviews: 2180 | Yes | | Terms score method | Lexicon based | Compared with online results | 84% | Xianghua et al., 2013 [26] | Reviews: 300 | Yes | | Terms score method | Lexicon based | Compared with tourist experts results | 75.89% | Wang et al., 2010 [23] | Reviews: 235,793 | | Yes | Support Vector Regression | Machine learning based | 5-fold cross-validation | 78% | Seki et al., 2009 [11] | Review: 1200 | Yes | | Naïve Bayes, SVM | Machine learning based | 3-fold cross-validation | 85% | Xueke et al., 2013 [27] | Reviews: 3214 | Yes | | Support Vector Machine | Machine learning based | 7-fold cross-validation | 83.9% | de Albornoz et al., 2011 [17] | Reviews: 1500 | Yes | | Logistic | Machine learning based | 3-fold cross-validation | 71.7% | Pontiki et al., 2014 [28] | Reviews: 300 | Yes | | Support Vector Machine | Machine learning based | 3-fold cross-validation | 80.15% | Pontiki et al., 2015 [29] | Reviews: 320 | Yes | | Maximum entropy | Machine learning based | 3-fold cross-validation | 78.69% | Proposed method | Reviews: 2000 (restaurants) | Yes | | FURIA | Machine learning based | 10-fold cross-validation | 90.12% | Proposed method | Reviews: 4000 (hotels) | Yes | | FLR | Machine learning based | 10-fold cross-validation | 86.02% |
|
|