4.2 Create Knowledge Base
In Pop, creating a knowledge base is very simple while still providing enough flexibility to adapt to different scenarios (such as personal notes, enterprise document repositories, private corpora, or intelligent customer‑service knowledge domains).
This section will guide you step by step through creating a knowledge base and understanding the purpose of each configuration item.
📌 1. Open the “Create Knowledge Base” Window
After entering the “Knowledge Base” module from the left navigation panel:
- Click the “＋ New Knowledge Base” button at the top
- Enter the “New Knowledge Base” creation form
- Fill in the basic information and choose an embedding model
The interface typically includes:
- Name input field
- Embedding model selection
- (Reserved) Knowledge base description
- Submit button
📌 2. Enter a Knowledge Base Name
The name identifies all content within the knowledge base.
Examples:
- “Company Policy Documents”
- “Technical Specifications”
- “Learning Notes Library”
- “Customer Service FAQ Base”
Naming rules:
| Type | Description |
|---|---|
| Length | Recommended within 30 characters |
| Duplicate names | Allowed; will not affect usage |
| Editable | Can be changed at any time after creation |
📌 3. Select an Embedding Model (Required)
The embedding model is one of the most important configurations, as it determines retrieval quality.
Pop currently supports three categories of embedding models:
① OpenAI Embedding Models (e.g., text-embedding-3-small)
- High precision
- Excellent multilingual performance
- Suitable for professional documents and multilingual corpora
② DeepSeek / Domestic Providers
- Lower cost
- Stronger performance for Chinese
- Lower latency
③ Local Models (Ollama / Self‑hosted Embedding Services)
- No data sent to cloud
- Suitable for private deployment
- Recommended for enterprise security scenarios
⚠️ Reminder: The embedding model cannot be modified once the knowledge base is created.
To change the model, you must create a new knowledge base.
📌 4. Knowledge Base Structure After Creation
After successful creation, you will see the main dashboard:
1. Document Management
Allows you to upload, view, and delete documents.
Supported formats include:
PDF, DOCX, PPTX, TXT, Markdown, Web pages, Code files, and more.
2. Indexing Status
Displays each document’s:
- Chunk count
- Embedding generation status
- Whether re‑indexing is required
- Error reports
3. Retrieval Interface
Provides three strategies:
- BM25 (keyword search)
- Vector search (KNN)
- Hybrid (BM25 + vector reranking)
4. Overview & Statistics
Includes:
- Total documents
- Total chunks
- Embedding model info
- Recent task history
📌 5. How to Rename the Knowledge Base
In the top‑right menu of the knowledge base detail page, you can:
- Edit the name
- View the knowledge base ID
- Delete the knowledge base
Deletion will require confirmation to prevent mistakes.
✅ Summary
Creating a knowledge base only requires two steps:
- Enter a name
- Select an embedding model