Skip to main content

Knowledge Base

📖 Best for: AI product managers / data teams / business operations — people who want to turn company documents, manuals, contracts, and training materials into an AI-searchable private knowledge base

📖 Reading time: 4 minutes

📖 In one sentence: YingCore's built-in RAG core module for building enterprise private knowledge assets, covering the full pipeline of create → upload → parse/chunk → configure → retrieval test → knowledge graph. Accessible from the top nav "知识库 / Knowledge Base" (route /knowledge) for all roles, powering workspace chat and intelligent apps to retrieve and cite.

Capabilities​

  • Multi-source ingestion: Upload PDF / Word / Excel / Markdown, crawl websites, sync databases, Feishu Docs, Yuque.
  • Auto-vectorization: Documents are auto-chunked, embedded, and stored in a vector database. Supports bge, bce, OpenAI text-embedding, and more.
  • Multimodal knowledge: Images, tables, and formulas are recognized and indexed—not just plain text.
  • Retrieval-augmented generation (RAG): Semantic, keyword, and hybrid retrieval with configurable Top-K, similarity threshold, and re-ranker.
  • Permission tiers: Set visibility by department or role, with secondary authentication for sensitive knowledge bases.
  • Versioning and rollback: Document changes are versioned; roll back to any historical version.
  • Metadata management: Tags, categories, expiration, owner—structured fields for filtering.

Typical Scenarios​

  • Product manual Q&A: Customer asks about specs → digital employee retrieves from knowledge base → returns accurate answer with source citations.
  • Onboarding: HR uploads policy, process, and benefits docs; new hires self-serve through conversation.
  • Legal contract search: Lawyer searches "contains bet-on条款 and signed in 2023" across hundreds of thousands of contracts—results in seconds.
  • Unified customer service scripts: All customer service digital employees share one knowledge base, ensuring consistent messaging.
  • Digital Employees — primary consumers of the knowledge base
  • Workflows — workflows often use the knowledge base as a retrieval node
  • Model Providers — embedding models are configured here
  • Wanxiang Graph — knowledge base can be combined with graph for entity-relation reasoning