Tweet Emotion Node
Emotion Recognition:
SemEval 2018 - Emotion Recognition (Mohammad et al., 2018) - 4 labels: anger, joy,sadness, optimism
Input
- payload string
- Tweet text (from twitter sensor)
Example message:
{
"payload": "@andst7 Important. South Korea is currently tracking to lose about half its population roughly every generation. Long lifespa…"
}Output
This node constitutes of multiple outputs which automatically get routed to the correct output based on classification.
| Output | Type |
|---|---|
| 1 | Anger |
| 2 | Joy |
| 3 | Optimism |
| 4 | Sadness |
The message will be enriched by the emotion property:
- payload string
- Original tweet text
- emotion.anger float
- Anger indicator; range=
0.0-1.0 - emotion.joy float
- Joy indicator; range=
0.0-1.0 - emotion.optimism float
- Optimism indicator; range=
0.0-1.0 - emotion.sadness float
- Sadness indicator; range=
0.0-1.0
Example message:
{
"payload": "@andst7 Important. South Korea is currently tracking to lose about half its population roughly every generation. Long lifespa…",
"emotion": {
"anger": 0.0289,
"joy": 0.0139,
"optimism": 0.0628,
"sadness": 0.0628
}
}Note, these values do not necessarily add up to 1.0.
Configuration
- Name optional
- unique name for this node
- Enable Filter flag
- Whether to enable threshold-based filtering
- Anger float [0.0-1.0]
- Specify the minimum required anger value
- Joy float [0.0-1.0]
- Specify the minimum required joy value
- Optimism float [0.0-1.0]
- Specify the minimum required optimism value
- Sadness float [0.0-1.0]
- Specify the minimum required sadness value
Higher values in thresholds can be used to increase certainty.
Last updated on
Sep 15, 2022