Local Knowledge

Local Knowledge (Document Storage)

What It Does

Local Knowledge creates a smart document database right in your browser. Think of it as a personal search engine for your documents - you can store articles, PDFs, notes, and then ask questions or search for specific information using natural language.

What Goes In, What Comes Out

Input

Name	Type	Description	Required	Default
database_name	Text	Name for your knowledge base	Yes	-
action	Text	What to do: create, insert, search, update, delete	Yes	-
document	Object	Document content and metadata	For insert	-
query	Object	Search query and filters	For search	-
embedding_dimension	Number	Vector size for search	No	768

Output

Name	Type	Description
success	Boolean	Whether the operation worked
results	Array	Found documents or operation results
statistics	Object	Database size and performance info
metadata	Object	Operation details and timing

Why Use Local Knowledge?

🔒 Complete Privacy: All documents stay in your browser, never uploaded anywhere 🌐 Works Offline: Search your documents even without internet 🔍 Smart Search: Find documents by meaning, not just keywords 💾 No Storage Limits: Store as many documents as your browser allows ⚡ Fast Search: Instant results from your local database

How It Works

flowchart LR
    A[📄 Your Documents] --> B[💾 Local Storage]
    B --> C[🔍 Smart Search]
    C --> D[📋 Relevant Results]

    style A fill:#e3f2fd
    style B fill:#fff3e0
    style C fill:#f3e5f5
    style D fill:#e8f5e8

Simple Process:

1. Store Documents: Add articles, notes, or any text content
2. Create Embeddings: Convert text to searchable vectors (automatic)
3. Search by Meaning: Ask questions or search with natural language
4. Get Smart Results: Find relevant documents even if they don’t contain exact keywords

Perfect For

📚 Personal Research: Store and search through research papers, articles, and notes 💼 Company Knowledge: Create searchable databases of policies, procedures, and documentation 🎓 Study Materials: Organize and search through educational content 📝 Content Collections: Build searchable libraries of any text content 🔍 Smart FAQ Systems: Create question-answering systems from your documents

Quick Setup Guide

1. Create a Knowledge Base

• Database Name: Choose a descriptive name like “research_docs” or “company_kb”
• Action: Set to “create” for first-time setup

2. Add Documents

• Action: Set to “insert”
• Document: Include title, content, and any metadata
• Embedding: Usually generated automatically by other nodes

3. Search Your Documents

• Action: Set to “search”
• Query: Ask questions or search terms
• Max Results: How many documents to return (5-10 is usually good)

Configuration Tips

For General Knowledge Base:

{
  "database_name": "my_knowledge_base",
  "action": "search",
  "similarity_threshold": 0.7,
  "max_results": 5
}

For Precise Technical Search:

{
  "database_name": "technical_docs",
  "action": "search",
  "similarity_threshold": 0.8,
  "max_results": 3
}

For Broad Research:

{
  "database_name": "research_collection",
  "action": "search",
  "similarity_threshold": 0.6,
  "max_results": 10
}

Browser Compatibility

Works in all major browsers:

• ✅ Chrome: Full support with large storage
• ✅ Firefox: Full support
• ⚠️ Safari: Limited large storage support
• ✅ Edge: Full support

Storage & Privacy

🔒 Complete Privacy: All documents stored locally in your browser 💾 Browser Storage: Uses IndexedDB for reliable, persistent storage 🔐 Optional Encryption: Can encrypt sensitive documents 🚫 No External Access: Data never leaves your device ♻️ Easy Cleanup: Delete knowledge bases anytime

Try It Yourself

Example 1: Personal Research Database

What you’ll build: Searchable collection of research articles and notes

Workflow:

Get All Text From Link → Recursive Character Text Splitter → Ollama Embeddings → Local Knowledge

Setup:

1. Create Database: Name it “research_db”
2. Add Documents: Insert articles with titles and content
3. Search: Ask questions like “What are the benefits of renewable energy?”

Result: Smart search through all your research materials.

Example 2: Company Knowledge Base

What you’ll build: Searchable company documentation and policies

Workflow:

Get HTML From Link → Local Knowledge → RAG Node → Edit Fields

Setup:

1. Build Collection: Add company docs, policies, procedures
2. Enable Search: Connect to RAG node for question-answering
3. Ask Questions: “What’s our vacation policy?” → Get accurate answers

Result: Smart company FAQ system with source citations.

Example 3: Study Materials Library

What you’ll build: Searchable collection of educational content

Workflow:

Upload Documents → Local Knowledge → Q&A Node → Download As File

Setup:

• Database Name: “study_materials”
• Content: Textbooks, lecture notes, articles
• Search: Find information by topic or concept

Result: Personal study assistant that finds relevant information instantly.

🔍 Advanced Example: Multi-Topic Knowledge Base

What you’ll build: Organized knowledge base with categories

Setup:

• Use metadata fields to categorize documents (topic, difficulty, source)
• Filter searches by category
• Create specialized collections for different subjects

Use case: Comprehensive learning system with organized, searchable content.

Best Practices

✅ Do This

• Use descriptive database names: “research_papers” vs “db1”
• Add meaningful metadata: Include titles, dates, categories
• Start with smaller collections: Test with 10-20 documents first
• Use appropriate similarity thresholds: 0.7 for general search, 0.8+ for precise matches
• Organize with metadata: Use categories and tags for better filtering

❌ Avoid This

• Storing duplicate documents (wastes space and confuses search)
• Using very low similarity thresholds (returns irrelevant results)
• Forgetting to include document titles and metadata
• Storing extremely long documents without splitting them first

Troubleshooting

💾 “Storage Quota Exceeded”

Problem: Can’t add more documents to knowledge base Solution: Delete old documents, enable compression, or request unlimited storage permission

🔍 “No Results Found”

Problem: Search doesn’t return any documents Solution: Lower similarity threshold to 0.6 or check if documents actually contain related content

🐌 Slow Search Performance

Problem: Searches take too long Solution: Reduce max_results, optimize database size, or use more specific search terms

📄 “Document Not Inserted”

Problem: Documents fail to save Solution: Check document format, ensure embeddings are included, verify database name is correct

Keeping Your Knowledge Base Organized

🧹 Simple Maintenance

• Remove old documents you no longer need
• Delete test databases when you’re done experimenting
• Keep related documents together for better search results

💡 Tips for Better Performance

• Start small with 10-20 documents, then add more
• Use clear document titles so you know what’s stored
• Group similar content (like all company policies in one database)

Related Nodes

🔗 Works Great With

• RAG Node: Uses Local Knowledge for document retrieval
• Q&A Node: Searches through stored documents for answers
• Recursive Character Text Splitter: Prepares documents for storage
• Ollama Embeddings: Creates searchable vectors from text

🔄 Required for RAG Workflows

Local Knowledge is essential for:

• Building smart Q&A systems
• Creating searchable document collections
• Enabling semantic search capabilities

What’s Next?

🌱 New to Document Storage?

1. Start Small: Create a test database with 5-10 documents
2. Add Content: Use Get All Text From Link to collect documents
3. Try Searching: Test different similarity thresholds
4. Build RAG System: Connect to RAG Node for smart Q&A

🚀 Ready for More?

• Explore RAG Node for intelligent document search
• Try Ollama Embeddings for creating document vectors
• Check out document processing examples

---

💡 Pro Tip: Start with a focused collection of related documents (like all your research papers on one topic) rather than mixing different types of content. This gives better search results and helps you understand how the system works.

Workflow Patterns

• AI-Powered Analysis Patterns
• Knowledge Base Integration
• Intelligent Content Processing

Related Tutorials

• Building Your First AI Workflow
• Advanced AI Integration

Practical Examples

• Real-World Use Cases
• Integration Examples
• Best Practice Examples

Related Nodes

Complementary Nodes

• RAGNode: Works well together in workflows
• QANode: Works well together in workflows
• RecursiveCharacterTextSplitter: Works well together in workflows
• OllamaEmbeddings: Works well together in workflows