sii_hybrid_search - Hybrid Search

Overview

The sii_hybrid_search tool performs intelligent hybrid search combining web search and cognitions from SII backend. Use separate queries for different sources: natural language for web search, keywords for cognition search. Features advanced relevance scoring, comprehensive filtering options, and intelligent result merging with source diversity.

Tool Name

• Internal Name: sii_hybrid_search
• Display Name: SII Hybrid Search
• Icon: Search

Prerequisites

1. SII Account: Valid SII account required
2. Environment Variables: SII_USERNAME, SII_PASSWORD, SII_BASE_URL (optional)

Parameters

Optional Parameters (at least one query required)

Parameter	Type	Default	Description
web_query	string	-	Natural language query for web search (e.g., "latest artificial intelligence developments"). If not provided, web search will be skipped.
cognition_query	string	-	Space-separated keywords for cognition search (e.g., "artificial intelligence development trends"). If not provided, cognition search will be skipped.
max_web_results	number	5	Maximum number of web search results to fetch (max: 10)
max_cognition_results	number	10	Maximum number of cognition results to fetch (max: 20)
result_limit	number	15	Maximum number of final merged results to return (max: 30)
cognition_sort	string	"hot"	Sort method for cognitions: "latest", "hot", "most_liked"
cognition_time_filter	string	"all"	Time filter for cognitions: "day", "week", "month", "all"
cognition_topics	string[]	-	Topics to filter cognitions by (e.g., ["AI", "Technology"])
cognition_source_filter	string	-	Source filter for cognitions (e.g., "arxiv", "reddit", "twitter")
cognition_author_filter	string	-	Author filter for cognitions

Features

1. Dual Search: Combines web search (latest online info) and cognition search (structured knowledge base)
2. Advanced Relevance Scoring: Content relevance, source diversity, intelligent ranking
3. Comprehensive Filtering: Time range, topic classification, source and author filters
4. Result Merging: Deduplication, relevance sorting, source diversity balancing

Usage Examples

Basic Hybrid Search

{
  "web_query": "latest artificial intelligence developments",
  "cognition_query": "artificial intelligence development trends"
}

Web Search Only

{
  "web_query": "2024 technology news",
  "max_web_results": 10
}

Cognition Search Only

{
  "cognition_query": "machine learning algorithms applications",
  "cognition_sort": "most_liked",
  "cognition_time_filter": "month"
}

Advanced Filtered Search

{
  "web_query": "latest deep learning research",
  "cognition_query": "deep learning neural networks",
  "cognition_topics": ["AI", "Machine Learning"],
  "cognition_source_filter": "arxiv",
  "cognition_time_filter": "week",
  "result_limit": 20
}

Search Strategies

Web Search Query

Use natural language, descriptive expressions:

• ✅ "latest applications of AI in healthcare"
• ✅ "how to manage state with React Hooks"
• ❌ "AI healthcare" (too brief)

Cognition Search Query

Use keywords, space-separated:

• ✅ "artificial intelligence healthcare applications cases"
• ✅ "React Hooks state management"
• ❌ "how to use React Hooks" (too conversational)

Best Practices

1. Combine Both Searches: Get comprehensive results balancing timeliness and depth
2. Targeted Queries: Use complete sentences for web, precise keywords for cognitions
3. Reasonable Result Counts: 5-10 for general tasks, 15-20 for deep research
4. Use Filters: Time filters for latest info, topic filters for specific domains
5. Check Result Quality: Review relevance scores and verify source reliability

Related Tools

• sii_web_search: Pure web search
• sii_cognitions: Pure cognition search
• sii_web_fetch: Fetch web content
• sii_deep_research: Deep research