TwitterX Sentiment

The Sentiment classifier is designed to identify and categorize the opinions expressed in a given text, especially for long text. Indeed, the classifier can act as a search filter. For instance, it provides the users the possibility to filter their search and focus on negative or positive content for further analysis. More use cases of the model would be available when used along with other classifiers. For instance, it could be used to detect the sentiment of the future action (when used in conjunction with the intent classifier), or could also be used to detect sentiment relating to organizations or people (when used in conjunction with the named entity recognition).

 

The Sentiment classifier can add one metadata section as an enrichment. The result is a ternary label (neutral, positive, negative) along with the predicted confidence level:

“enrichment”:
{
“sentiment”: {
“label”: “True”,
“confidence”: 0.95
},
}