Classify texts by SDG using ChatGPT.
The prompt does not include examples for each SDG. It relies on ChatGPT's existing knowledge of the SDGs.
Response is in JSON format.
model
: ChatGPT model to use
The table below shows the accuracy (in %) for all configurations.
Configuration | Average | SDG 1 | SDG 2 | SDG 3 | SDG 4 | SDG 5 | SDG 6 | SDG 7 | SDG 10 |
---|---|---|---|---|---|---|---|---|---|
1 | 87.3 | 76.6 | 89.9 | 82.9 | 85.4 | 88.4 | 91.8 | 90 | 93.4 |
2 | 83.1 | 61 | 89.9 | 80.3 | 82.9 | 84.1 | 90.6 | 91 | 85.2 |
Parameters:
model
: gpt-4-0125-preview
SDG | n | Accuracy (%) | Precision (%) | Recall (%) | F1 Score | TP | FP | TN | FN |
---|---|---|---|---|---|---|---|---|---|
Average | 77.4 | 87.3 | 80.7 | 97.4 | 0.88 | 37.2 | 9 | 30.2 | 0.9 |
1 | 77 | 76.6 | 60.5 | 96.3 | 0.74 | 26 | 17 | 33 | 1 |
2 | 69 | 89.9 | 89.6 | 95.6 | 0.92 | 43 | 5 | 19 | 2 |
3 | 76 | 82.9 | 68.3 | 100.0 | 0.81 | 28 | 13 | 35 | 0 |
4 | 82 | 85.4 | 79.2 | 97.7 | 0.88 | 42 | 11 | 28 | 1 |
5 | 69 | 88.4 | 81.4 | 100.0 | 0.90 | 35 | 8 | 26 | 0 |
6 | 85 | 91.8 | 87.3 | 100.0 | 0.93 | 48 | 7 | 30 | 0 |
7 | 100 | 90.0 | 83.3 | 100.0 | 0.91 | 50 | 10 | 40 | 0 |
10 | 61 | 93.4 | 96.3 | 89.7 | 0.93 | 26 | 1 | 31 | 3 |
Evaluated on: May 10, 2024
Parameters:
model
: gpt-3.5-turbo-0125
SDG | n | Accuracy (%) | Precision (%) | Recall (%) | F1 Score | TP | FP | TN | FN |
---|---|---|---|---|---|---|---|---|---|
Average | 77.4 | 83.1 | 76.3 | 97.1 | 0.85 | 37 | 11.8 | 27.5 | 1.1 |
1 | 77 | 61.0 | 47.4 | 100.0 | 0.64 | 27 | 30 | 20 | 0 |
2 | 69 | 89.9 | 95.2 | 88.9 | 0.92 | 40 | 2 | 22 | 5 |
3 | 76 | 80.3 | 65.9 | 96.4 | 0.78 | 27 | 14 | 34 | 1 |
4 | 82 | 82.9 | 76.4 | 97.7 | 0.86 | 42 | 13 | 26 | 1 |
5 | 69 | 84.1 | 77.3 | 97.1 | 0.86 | 34 | 10 | 24 | 1 |
6 | 85 | 90.6 | 85.7 | 100.0 | 0.92 | 48 | 8 | 29 | 0 |
7 | 100 | 91.0 | 84.7 | 100.0 | 0.92 | 50 | 9 | 41 | 0 |
10 | 61 | 85.2 | 77.8 | 96.6 | 0.86 | 28 | 8 | 24 | 1 |
Evaluated on: May 10, 2024