|
--- |
|
language: |
|
- en |
|
tags: |
|
- sentiment-analysis |
|
- sklearn |
|
--- |
|
|
|
## Overview |
|
Sentinet V1 is a collection of models to thoroughly analyze the sentiments, emotions of a given text. |
|
|
|
The underlying algorithm is TF-IDF Vectorization followed by Logistic Regression |
|
|
|
## Performance |
|
sentiment_class | auroc_score |
|
---|---: |
|
sentiment_polarity | 95.04% |
|
opinion | 70.64% |
|
toxicity | 96.12% |
|
toxicity__hate | 97.43% |
|
toxicity__insult | 97.04% |
|
toxicity__obscene | 98.44% |
|
toxicity__sexual_explicit | 98.49% |
|
toxicity__threat | 98.25% |
|
emotion__anger | 86.36% |
|
emotion__disgust | 85.15% |
|
emotion__fear | 93.03% |
|
emotion__guilt | 81.70% |
|
emotion__humour | 97.69% |
|
emotion__joy | 85.87% |
|
emotion__no_emotion | 80.08% |
|
emotion__sadness | 91.04% |
|
emotion__shame | 84.19% |
|
emotion__surprise | 87.55% |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
