Comparing Spider results with Peaka AI to show how semantics boost SQL accuracy.
Tool/Library | Purpose |
---|---|
Peaka | A zero-ETL data integration platform with single step context generation capability |
Spider Dataset | Source of natural language questions and structured DB schemas |
Snowflake SQL | Backend execution engine used by Spider for SQL result generation |
Requests | Used to make authenticated API calls to Peaka AI Agent |
Streamlit | To build an interactive web UI for comparing question results |
TO_NUMBER
→ CAST AS DECIMAL
REGEXP_EXTRACT
→ compatible Trino equivalentGET_URL: Fetches info about your Peaka project (including projectId).
POST_URL: Sends questions to the AI agent in that project.Headers include your token for authorization: