Table 3.
Examples of feedback instances from our dataset along with their semantic similarity scores calculated using pretrained ‘all-MiniLM-L12-v2’ sentence-embeddings (SBERT)42
| Feedback Instance Example | Feedback for Comparison | Similarity Score |
|---|---|---|
| “okay, that’s pretty good” | “yeah, that’s fine” | 0.506 |
| “closer and rotate” | 0.119 | |
| “so burn... coag there first” | 0.121 | |
| “closer and rotate” | “turn don’t push, turn” | 0.377 |
| “yeah, that’s fine” | 0.058 | |
| “this is a lot of bleeding” | 0.042 | |
| “so burn... coag there first” | “this is a lot of bleeding” | 0.257 |
| “yeah, that’s fine” | 0.056 | |
| “closer and rotate” | 0.094 |