Configuration Guide

Semantic Layer Development

Goal: Help the system correctly understand data and return stable, accurate answers

Step 1 — Plan analysis domains

Analysis Domain Planning Guide — Enterprise-level analysis domain division methodology and best practices

Step 2 — Connect data and models

Data Source Management — Add Lakehouse, MySQL, StarRocks, Databricks, and others Model Selection and Configuration — Configure the LLM model used for Q&A

Step 3 — Build the analysis domain

Configure Analysis Domain — Create analysis domains, add tables, configure table details and permissions

Step 4 — Configure field semantics

Configure Field Semantics — Aliases, descriptions, types, usage, hidden, indexes Configure Virtual Columns — Create derived fields

Step 5 — Configure definitions and knowledge

Metrics and Answer Builder — Create metrics and Answer Builders to lock in calculation definitions Configure Knowledge — Configure business definitions and terminology explanations File and Document Q&A — Upload documents for document-based Q&A

Operations and Governance

Goal: Ensure the system is secure, governable, and auditable

Permissions and Security

Row-Level Permissions — Control the data range accessible to users Bulk Download and Data Export Governance — Manage download permissions, export auditing

Operations

Web Search — Enable external web retrieval for an analysis domain Message Notifications — View background task status for file imports, table imports, etc.

Q&A Quality

Troubleshoot Q&A Accuracy Issues — Complete investigation from answer to SQL to records Launch Checklist for Analysis Domain — Health check and pre-launch checklist Validate Q&A Quality — Validate whether configurations are effective with typical questions Handle Feedback — Admin-side viewing and assignment of user feedback