surface↓classification↓stance
Classificationdimension · returns varchar

STANCE

Identify stance on a topic via zero-shot NLI (topic required)

Per-row classifier — stable across GROUP BY.

classificationnlispecialist-zootext

Arguments

nametypedescription
textVARCHAR—
topicVARCHARThe topic to analyze stance on
stancesVARCHAROptional custom stance labels (comma-separated)

About

Stance analyzer — identifies the author's stance on a specific topic. **REQUIRES** a topic argument. Backend: specialist zoo zero-shot NLI via /classify_batch. The hypothesis template embeds the topic directly so the NLI model scores each text against stance labels in the context of that topic. Default stances are {supportive of, opposed to, neutral toward, mixed about, off-topic on}. Pass a custom comma-separated `stances` argument to override. Usage: SELECT stance(review, 'AI regulation'), COUNT(*) FROM articles GROUP BY stance(review, 'AI regulation') For LLM-style stance detection with rhetorical nuance, use STANCE_LLM — see stance_dimension_llm.cascade.yaml.

Examples

Stance detected

SELECT
  stance (
    'I strongly support this initiative',
    'the initiative'
  )

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