snowflake.snowpark.DataFrameAIFunctions.sentiment

DataFrameAIFunctions.sentiment(input_column: Union[snowflake.snowpark.column.Column, str], categories: Optional[List[str]] = None, *, output_column: Optional[str] = None) → snowflake.snowpark.DataFrame[source]

Extract sentiment analysis from text content.

This method analyzes the sentiment of text in each row, providing overall sentiment and optionally sentiment for specific categories or aspects mentioned in the text.

Parameters:

  • input_column – The column (Column object or column name as string) containing the text to analyze for sentiment.

  • categories – Optional list of up to 10 categories (also called entities or aspects) for which sentiment should be extracted. Each category may be a maximum of 30 characters long. For example, if extracting sentiment from restaurant reviews, you might specify ['cost', 'quality', 'service', 'wait time'] as categories. If not provided, only overall sentiment is returned.

  • output_column – The name of the output column to be appended. If not provided, a column named AI_SENTIMENT_OUTPUT is appended.

Returns:

A new DataFrame with an appended output column containing sentiment results. The output is a JSON object with a categories field containing an array of records. Each record includes:

  • name: The category name (overall for overall sentiment)

  • sentiment: One of unknown, positive, negative, neutral, or mixed

Examples:

>>> # Overall sentiment analysis
>>> df = session.create_dataframe([
...     ["The movie had amazing visual effects but the plot was terrible."],
...     ["The food was delicious but the service was slow."],
...     ["Everything about this experience was perfect!"],
... ], schema=["review"])
>>> result_df = df.ai.sentiment(
...     input_column="review",
...     output_column="sentiment"
... )
>>> result_df.columns
['REVIEW', 'SENTIMENT']
>>> results = result_df.collect()
>>> import json
>>> overall_sentiment = json.loads(results[2]["SENTIMENT"])["categories"][0]
>>> overall_sentiment["name"]
'overall'
>>> overall_sentiment["sentiment"]
'positive'

>>> # Sentiment analysis with specific categories
>>> from snowflake.snowpark.functions import col
>>> df = session.create_dataframe([
...     ["The hotel room was spacious and clean, but the wifi was terrible and the breakfast was mediocre."],
...     ["Great location and friendly staff, though the parking was expensive."],
... ], schema=["review"])
>>> result_df = df.ai.sentiment(
...     input_column=col("review"),
...     categories=["room", "wifi", "breakfast", "location", "staff", "parking"],
...     output_column="detailed_sentiment"
... )
>>> result_df.columns
['REVIEW', 'DETAILED_SENTIMENT']
>>> results = result_df.collect()
>>> sentiments = json.loads(results[0]["DETAILED_SENTIMENT"])["categories"]
>>> # Check that we have sentiments for overall plus the specified categories
>>> len(sentiments) > 1
True
>>> category_names = [s["name"] for s in sentiments]
>>> "overall" in category_names
True
>>> "room" in category_names
True

Note

AI_SENTIMENT can analyze sentiment in English, French, German, Hindi, Italian, Spanish, and Portuguese. You can specify categories in the language of the text or in English.

This function or method is experimental since 1.39.0.