Skip to main content

​
Overview

Counts brand and product name mentions in the transcript. Requires at least one configuration parameter - Provide brand_name or product_name.

​
Configuration

​
config.brand_name
string
Brand name to count.
​
config.product_name
string
Product name to count.

​
Example Configuration

{
  "detector_name": "transcript_product",
  "config": {
    "brand_name": "Acme",
    "product_name": "Turbo Blender"
  }
}

​
Result Schema

{
  "detector_name": "transcript_product",
  "pass_check": true,
  "score": 1.0,
  "rationale": "All names meet minimum: 'Acme': 2 mentions, 'Turbo Blender': 1 mentions",
  "metrics": {
    "exact_count": 3,
    "brand_count": 2,
    "product_count": 1,
    "partial_count": 0,
    "out_of_order_count": 0,
    "matches": {
      "exact": [[5, 7]],
      "brand": [[5, 6]],
      "product": [[6, 7]],
      "partial": [],
      "out_of_order": []
    },
    "matches_char": {
      "exact": [{"start": 42, "end": 54, "text": "Acme Turbo"}],
      "brand": [{"start": 42, "end": 46, "text": "Acme"}],
      "product": [{"start": 47, "end": 54, "text": "Turbo"}],
      "partial": [],
      "out_of_order": []
    },
    "brand_name": "Acme",
    "product_name": "Turbo Blender",
    "min_mentions": 1,
    "count_mode": "separate"
  }
}

​
Interpreting Results

  • brand_count/product_count: Mention counts per name.
  • matches: Token-span matches by type (exact/brand/product).
  • matches_char: Character-span evidence for each match.