Methodology

Each article goes through a two-pass LLM rating pipeline using GPT-5.4-nano (reasoning_effort=low, max_tokens=2000).

Pass 1: Relevance Filter

Determines if the article is ACTUALLY about the dog breed, or just mentions it incidentally.

Loading...

Pass 2: Breed Sentiment Rating (1-10)

For relevant articles only — rates how positive or negative the public sentiment toward the breed is as conveyed in the article.

Loading...