Principle:Infiniflow Ragflow Chat Application Creation
| Knowledge Sources | |
|---|---|
| Domains | RAG, Conversational_AI |
| Last Updated | 2026-02-12 06:00 GMT |
Overview
A configuration pattern that creates a RAG-powered chat application by binding knowledge bases, an LLM, and retrieval parameters into a reusable dialog entity.
Description
Chat Application Creation establishes a Dialog entity that encapsulates all configuration needed for RAG-powered conversations: linked knowledge bases, LLM model selection, generation parameters (temperature, top_p, penalties), retrieval parameters (similarity threshold, vector weight, top_n), and prompt templates (system prompt, prologue, parameters, empty response). This entity serves as the reusable configuration for all conversations created under it.
Usage
Create a chat application after knowledge bases have been populated with processed documents. Each chat application can link to multiple knowledge bases and be configured with different LLM and retrieval settings.
Theoretical Basis
The chat application pattern separates configuration from execution:
- Configuration (Dialog): Static settings defining how RAG operates
- Execution (Conversation): Stateful session tracking message history
- This separation allows multiple concurrent conversations to share the same configuration