Table 1 Model accuracy using different embedding techniques
% | AvgVec | Doc2Vec | NNLM | USE | BERT | AL- BERT | ELEC- TRA |
|---|---|---|---|---|---|---|---|
Actor | 66.47% | 72.61% | 67.44% | 72.67% | 56.77% | 50.96% | 48.67% |
Agency | 92.65% | 78.61% | 69.35% | 72.99% | 94.78% | 94.31% | 95.43% |
SDG | 66.49% | 58.61% | 55.98% | 57.38% | 61.84% | 59.93% | 61.74% |
Sector | 55.70% | 49.67% | 48.97% | 53.14% | 55.19% | 55.61% | 56.66% |
Place | 79.41% | 76.66% | 72.32% | 88.39% | 84.70% | 82.53% | 73.95% |
Year | 74.01% | 55.62% | 34.73% | 51.04% | 83.81% | 78.72% | 80.81% |
Month | 63.84% | 40.15% | 22.93% | 34.28% | 70.41% | 66.22% | 69.62% |
Reason | 69.60% | 57.97% | 56.06% | 61.11% | 70.95% | 71.31% | 70.87% |
Average | 71.02% | 61.24% | 53.47% | 61.37% | 72.31% | 69.95% | 69.72% |
Time | 0H:58M | 1H:12M | 1H:24M | 6H:28M | 15H:18M | 10H:47M | 14H:43M |