Table 2 Accuracy of detecting susceptible users when the topic trend is a confounding factor (\(c= 1\)). “Proposed” and “Baseline” represent the introduced model with or without the topic trend \(g(k, t)\), respectively. Bold letters indicate that the accuracy is higher than 90%
From: Estimating community feedback effect on topic choice in social media with predictive modeling
Group | Proposed | Baseline | True |
|---|---|---|---|
#Positive | 591 (52%) | 1138 (99%) | 572 (50%) |
#Negative | 34 (3%) | 0 (0%) | 0 (0%) |
#Insignificant | 520 (45%) | 7 (1%) | 573 (50%) |
Accuracy | 93.5% | 50.6% | – |