Query Fan-Out Content Outline Generator

PromptCopy prompt

`text
## Comprehensive Query Fan-Out Prompt: Patent-Aligned Algorithm Implementation


### Understanding Query Fan-Out: The Technical Framework


Query fan-out is an advanced multi-stage process in which search engines break down a user's query through several algorithmic steps:


### Input Requirements
1. **Primary Query**: {primary_query} 
2. **User Context**: {user_context}
3. **Business Objective**: {business_objective} 


1. **Initial Query Decomposition**: The system pinpoints the core semantic components and entities inside the query
2. **Thematic Clustering**: Components are grouped into separate themes based on semantic relationships
3. **Hierarchical Expansion**: Each theme spawns sub-themes and related concepts in a tree structure
4. **Contextual Rewriting**: Queries are reworded according to user context and implicit intent 
5. **Entity Disambiguation**: The system links query components to Knowledge Graph entities
6. **Ranking and Filtering**: Generated queries are scored and filtered using relevance signals


### Detailed Query Fan-Out Algorithm


#### Stage 1: Query Decomposition & Entity Recognition


**Split the query into semantic components:**
- **Entities**: Identify every named entity (people, places, products, concepts)
- **Attributes**: Pull out descriptive properties (size, color, quality, temporal)
- **Relationships**: Map the connections between entities (comparisons, dependencies, hierarchies)
- **Actions/Intents**: Identify verbs and implied actions


Example: "best sustainable marketing strategies for small e-commerce"
- Entities: [marketing strategies, e-commerce]
- Attributes: [best, sustainable, small]
- Relationships: [strategies FOR e-commerce]
- Intent: [evaluation/selection]


#### Stage 2: Thematic Clustering with Semantic Boundaries


**Produce 3-5 primary themes where each theme:**
- Represents a separate informational facet
- Has clear semantic boundaries (what's included/excluded)
- Contains a "centroid" concept (most representative element)
- Includes "peripheral" concepts (boundary elements)


**Theme Generation Rules:**
1. Every theme must serve a different user need
2. Themes should share 30-70% semantic overlap with the seed query
3. Build both "convergent" themes (narrowing focus) and "divergent" themes (broadening scope)


#### Stage 3: Hierarchical Query Tree Construction


For each theme, assemble a query tree with these levels:


**Level 1 - Root Queries** (Direct theme representation)
- Broadest form of the theme
- Example: "sustainable marketing methods"


**Level 2 - Aspect Queries** (Specific facets)
- Split the root into 3-4 separate aspects
- Example: "environmental impact of digital marketing"


**Level 3 - Detail Queries** (Granular questions)
- Specific, actionable queries
- Example: "carbon footprint of email campaigns calculation"


**Level 4 - Context Queries** (Situational variations)
- Adjusted by user context
- Example: "email carbon footprint for B2B SaaS companies"


#### Stage 4: Query Rewriting Mechanisms


Apply these rewriting patterns to every base query:


**1. Lexical Substitution**
- Swap terms for synonyms/related terms
- "strategies" → "tactics", "approaches", "methods"


**2. Structural Reformulation**
- Statement → Question: "sustainable marketing" → "what makes marketing sustainable?"
- Broad → Specific: "marketing strategies" → "content marketing strategies"
- Abstract → Concrete: "best practices" → "step-by-step guide"


**3. Intent Transformation**
- Information → Action: "what is X" → "how to implement X"
- General → Comparative: "good strategies" → "X vs Y strategies"
- Single → Multiple: "best strategy" → "top 5 strategies"


**4. Contextual Augmentation**
- Add temporal context: "...in 2024", "...post-COVID"
- Add expertise level: "...for beginners", "...advanced techniques"
- Add industry specific: "...for SaaS", "...in healthcare"


#### Stage 5: Entity-Based Expansion


**Knowledge Graph Integration:**
1. Link each entity to its knowledge graph node
2. Retrieve related entities (parent, child, sibling relationships)
3. Extract entity attributes and properties
4. Generate queries that combine entities and attributes


**Entity Relationship Patterns:**
- **Hierarchical**: broader/narrower terms
- **Associative**: commonly co-occurring entities
- **Comparative**: similar entities for comparison
- **Compositional**: part-whole relationships


#### Stage 6: Scoring and Ranking


**Apply multi-factor scoring to each generated query:**

Copy to clipboardDownload .md