Table 4 Results (%) of the hybrid method (Logistic Regression + IR) on community QA datasets (SemEval-cQA-Test 2016 and SemEval-cQA-Test 2017)
Systems | Test Sets | Acc | P | R | F1 | MAP | MRR |
|---|---|---|---|---|---|---|---|
Hybrid Method (Logistic Regression + IR) | cQA-16-Test | 80.57 | 70.29 | 72.10 | 71.19 | 77.47 | 83.79 |
cQA-B-2016 Best System [ 4 ] | cQA-16-Test | 76.57 | 63.53 | 69.53 | 66.39 | 76.70 | 83.02 |
cQA-B-2016 IR Baseline [ 4 ] | cQA-16-Test | - | - | - | - | 74.75 | 83.79 |
Hybrid Method (Logistic Regression + IR) | cQA-17-Test | 67.27 | 33.68 | 79.14 | 47.25 | 44.66 | 48.08 |
cQA-B-2017 Best System [ 5 ] | cQA-17-Test | 52.39 | 27.30 | 94.48 | 42.37 | 47.22 | 50.07 |
cQA-B-2017 IR Baseline [ 5 ] | cQA-17-Test | - | - | - | - | 41.85 | 46.42 |