Skip to main content

​
Overview

Evaluates sentiment and prohibited content in the transcript. Requires at least one configuration parameter - Provide desired_sentiment, undesired_sentiment, or prohibited_content_types.

​
Configuration

​
config.desired_sentiment
string
Desired sentiment or tone to detect.
​
config.undesired_sentiment
string
Undesired sentiment or tone to avoid.
​
config.prohibited_content_types
array
List of prohibited content types to detect.

​
Prohibited Content Types

  # Illegal/harmful content
  "illegal_conduct",
  "drugs_alcohol",
  "violence",
  "abuse",
  "self_harm",
  "harm_others",
  # Hate & harassment
  "racism",
  "sexism",
  "homophobia",
  "transphobia",
  "discrimination",
  "harassment",
  "bullying",
  # Sexual/adult content
  "pornography",
  "sexually_explicit",

​
Example Configuration

{
  "detector_name": "sentiment",
  "config": {
    "desired_sentiment": "optimistic and upbeat",
    "undesired_sentiment": "hostile",
    "prohibited_content_types": ["bullying", "hate_speech"]
  }
}

​
Result Schema

{
  "detector_name": "sentiment",
  "pass_check": true,
  "score": 0.67,
  "rationale": "Sentiment criteria met",
  "metrics": {
    "desired_sentiment": "optimistic and upbeat",
    "undesired_sentiment": "hostile",
    "prohibited_content_types": ["bullying", "hate_speech"],
    "desired_count": 2,
    "undesired_count": 0,
    "prohibited_count": 0,
    "desired_matches": [{"text": "great day", "explanation": "positive phrasing"}],
    "undesired_matches": [],
    "prohibited_content_matches": [],
  }
}