surface↓classification↓sentiment_llm
Classificationdimension · returns varchar

SENTIMENT_LLM

LLM-backed dimensional sentiment analysis (escape hatch for SENTIMENT)

Per-row classifier — stable across GROUP BY.

classificationllmllm-escape-hatchtext

Arguments

nametypedescription
textVARCHAR—
focusVARCHAR—
num_levelsINTEGER—

About

LLM-backed alternative to SENTIMENT for sentiment classification when the base model's signal misses nuance — sarcasm, mixed feelings, domain-specific emotional tone, or sentiment-bearing language that's unusual in the training data (technical complaints, politeness-coded criticism, culturally specific expressions). Useful on support tickets, reviews of niche products, or subjective feedback where the tone matters more than the surface words. For the vast majority of sentiment work, SENTIMENT is dramatically cheaper and runs a purpose-built classifier. Reach for the LLM variant selectively on rows where you've already flagged potential nuance.

Examples

LLM escape hatch assesses sentiment

SELECT
  sentiment_llm ('I had a great experience')

Nearby rabbit holes

same domain
dimension

AUDIENCE

Identify target audience via zero-shot NLI

classificationnlispecialist-zootext+1
dimension

AUDIENCE_LLM

LLM-backed audience identification (escape hatch for AUDIENCE)

classificationllmllm-escape-hatchtext
dimension

AUTHENTICITY

Assess authenticity of content via zero-shot NLI

classificationnlispecialist-zootext
dimension

AUTHENTICITY_LLM

LLM-backed authenticity assessment (escape hatch for AUTHENTICITY)

classificationllmllm-escape-hatchtext
dimension

BUCKET

Classify text into user-specified buckets via zero-shot NLI

classificationnlispecialist-zootext
dimension

BUCKET_LLM

LLM-backed bucketing (escape hatch for BUCKET)

classificationllmllm-escape-hatchtext
Climb back to The Looking Glass