Categories:: String & binary functions (AI Functions)

SENTIMENT (SNOWFLAKE.CORTEX)¶

Returns an overall sentiment score for the given English-language input text.

Syntax¶

SNOWFLAKE.CORTEX.SENTIMENT(<text>)

Copy

Arguments¶

text: A string containing the text for which a sentiment score should be calculated.

Returns¶

A floating-point number from -1 to 1 (inclusive) indicating the model’s level of certainty of any detected sentiment. A score close to 0 indicates that the function could not determine a clear sentiment in the text; this result can be considered neutral. A score close to 1 indicates positive sentiment, while a score close to -1 indicates negative sentiment. The chart below provides guidance on how to interpret the sentiment scores:

Sentiment	Sentiment Score
Positive	0.5 to 1
Neutral	-0.5 to 0.5
Negative	-0.5 to -1

The result does not indicate the intensity of sentiment, but the polarity (positive, neutral, or negative) and certainty.

Access control requirements¶

Users must use a role that has been granted the SNOWFLAKE.CORTEX_USER database role. See Cortex LLM privileges for more information on this privilege.

Examples¶

The following example uses SENTIMENT to get the sentiment classification of a food service review, which we can infer as modestly positive, given the score of 0.54.

SELECT SNOWFLAKE.CORTEX.SENTIMENT('A tourist\'s delight, in low urban light,
  Recommended gem, a pizza night sight. Swift arrival, a pleasure so right,
  Yet, pockets felt lighter, a slight pricey bite. 💰🍕🚀');

Copy

Response:

0.5424458

In the following example, a table named reviews contains a column named review_content containing the text of reviews submitted by users. The query returns a sentiment score for each review.

SELECT SNOWFLAKE.CORTEX.SENTIMENT(review_content), review_content FROM reviews LIMIT 10;

Copy

Legal notices¶

Refer to Snowflake AI and ML.

Limitations¶

Snowflake Cortex functions do not support dynamic tables.