Basic Documents — Sentiment Fields

The sentiment reflects the emotional tone of the underlying document, from optimistic ("positive") to neutral ("neutral") to pessimistic ("negative").

Field NameTypeExample/Possible ValuesDescription
sentiment_predictionstringpositive, neutral, or negativeThe sentiment prediction of the document.
positive_sentiment_probabilitynumber0 ≤ value ≤ 1The probability of the document being positive.
neutral_sentiment_probabilitynumber0 ≤ value ≤ 1The probability of the document being neutral.
negative_sentiment_probabilitynumber0 ≤ value ≤ 1The probability of the document being negative.
sentiment_extremitynumber0 ≤ value ≤ 1The sentiment extremity of the document.

sentiment_prediction

Definition: Each data row is assigned a sentiment label (one of positive, neutral or negative), reflecting the overall tone of the underlying document.

Possible values: positive, neutral, or negative.

positive_sentiment_probability

Definition: The probability that the underlying document has positive sentiment.

Possible values: A float in the interval [0, 1] (i.e., between 0 and 1, inclusive).

Note: The three values positive_sentiment_probability, negative_sentiment_probability and neutral_sentiment_probability always sum to 1.0.

neutral_sentiment_probability

Definition: The probability that the underlying document has neutral sentiment.

Possible values: A float in the interval [0, 1] (i.e., between 0 and 1, inclusive).

Note: The three values positive_sentiment_probability, negative_sentiment_probability and neutral_sentiment_probability always sum to 1.0.

negative_sentiment_probability

Definition: The probability that the underlying document has negative sentiment.

Possible values: A float in the interval [0, 1] (i.e., between 0 and 1, inclusive).

Note: The three values positive_sentiment_probability, negative_sentiment_probability and neutral_sentiment_probability always sum to 1.0.

sentiment_extremity

Definition: A metric that reflects the extremity of the discrete probability distribution:

(positive_sentiment_probability , negative_sentiment_probability , neutral_sentiment_probability).

Here, "extremity" means the degree to which this distribution is both non-neutral, and has widely differing positive and negative components. More precisely, this metric has two key properties:

  1. If either of the fields positive_sentiment_extremity or negative_sentiment_extremity has high value, then the sentiment_extremity will be close to 1.0.
  2. If either the field neutral_sentiment_extremity has high value, or if the fields positive_sentiment_extremity and negative_sentiment_extremity have similar values, then the sentiment_extremity will be close to 0.0.

Possible values: A float in the interval [0, 1] (i.e., between 0 and 1, inclusive).