Research Article
Fast Detection of Deceptive Reviews by Combining the Time Series and Machine Learning
Table 4
Effect results for 11 features.
| | Features | Hit rate | Recall | F1 measures | Precision |
| | Duplicates reviews | 0.58 | 0.40 | 0.48 | 0.49 | | Self-reference | 0.83 | 0.18 | 0.30 | 0.51 | | The length of review | 0.79 | 0.80 | 0.79 | 0.76 | | The ratio of opinion words | 0.81 | 0.47 | 0.60 | 0.61 | | Transition words of the sentiment expression | 0.72 | 0.15 | 0.25 | 0.48 | | Exclamation tone | 0.85 | 0.18 | 0.30 | 0.51 | | Appending time | 0.63 | 0.80 | 0.70 | 0.55 | | Appending pictures | 0.58 | 0.90 | 0.70 | 0.57 | | User activity | 0.59 | 0.85 | 0.69 | 0.57 | | Super member | 0.57 | 0.42 | 0.48 | 0.48 | | Review posting | 0.58 | 0.92 | 0.71 | 0.59 |
|
|