Sentiment analysis prebuilt model
The sentiment analysis prebuilt model detects positive or negative sentiment in text data. You can use it to analyze social media, customer reviews, or any text data you're interested in. Sentiment analysis evaluates text input, and gives scores and labels at a sentence and document level. The scores and labels can be positive, negative, or neutral. At the document level, there can also be a "mixed" sentiment label, which has no score. The sentiment of the document is determined by aggregating the sentence scores.
Use in Power Apps
Explore sentiment analysis
You can try out the sentiment analysis model before you import it into your flow by using the "try it out" feature.
Sign in to Power Apps.
In the left pane, select AI Builder > Build.
Under Get straight to productivity, select Sentiment Analysis.
In the Sentiment Analysis window, select Try it out.
Select predefined text samples to analyze, or add your own text in the Add your own here box to see how the model analyzes your text.
Use the formula bar
You can integrate your AI Builder sentiment analysis models in Power Apps Studio by using the formula bar. For more information, see Use formulas for text AI models.
Use in Power Automate
If you want to use this prebuilt model in Power Automate, you can find more information in Use the sentiment analysis prebuilt model in Power Automate.
Supported language and data format
- Language: German, Spanish, English, French, Hindi, Italian, Japanese, Korean, Dutch, Norwegian, Portuguese (Brazil), Portuguese (Portugal), Turkish, Chinese (Simplified), Chinese (Traditional)
- Documents can't exceed 5,120 characters.
If text is detected, the sentiment analysis model outputs the following information:
DocumentScores: Value in the range from 0 through 1. Values close to 1 indicate greater confidence that the identified sentiment is accurate.
Sentences: List of sentences from the input text, with analysis of its sentiments.
SentenceScores: Value in the range from 0 through 1. Values close to 1 indicate greater confidence that the sentiment is accurate.
The following applies to calls made per environment across the following prebuilt models: language detection, sentiment analysis, and key phrase extraction.
|Calls (per environment)||400||60 seconds|