Knowledge Graph - Multi Hop Reasoning Query Builder

Generate graph queries that perform multi-hop traversal and reasoning across relationships in graph databases and knowledge graphs.

Multi-Hop Reasoning Query Builder

Construct graph queries that explore indirect relationships through multiple hops in a graph.

This skill helps developers generate queries that traverse several relationships to discover hidden connections and complex patterns within graph datasets.

Multi-hop reasoning is commonly used in:

• knowledge graph exploration
• recommendation systems
• fraud detection
• supply chain analysis
• social network analysis

The skill produces queries that traverse multiple edges to identify connections between entities.

---

📋 Quick Start

When To Use This Skill

Use this skill when a user wants to:

• build multi-hop graph queries
• explore indirect relationships
• perform reasoning across graph paths
• discover connections through several relationships
• construct path traversal queries
• find hidden patterns across hops
• analyze relationship chains

Example Requests

• "Find friends of friends of Alice."
• "Find suppliers connected to a company within three hops."
• "Discover customers linked through purchase chains."
• "Find entities connected to a node within multiple relationships."
• "Show me the path from Product A to Product B."
• "Find all paths between two companies."

---

🎯 What This Skill Produces

The skill generates graph queries capable of:

• Multi-hop traversal - Traverse multiple relationship hops
• Path discovery - Find paths between entities
• Indirect relationship detection - Discover hidden connections
• Graph reasoning queries - Complex reasoning patterns
• Variable-length path exploration - Flexible traversal patterns
• Performance-optimized queries - Efficient multi-hop queries

Outputs include:

• Cypher traversal queries
• SPARQL property path queries
• Graph path exploration queries
• Parameterized multi-hop templates

---

🔍 Multi-Hop Query Types

1. Fixed Depth Multi-Hop

Pattern: Traverse exactly N relationships

Cypher:

cypher

MATCH (start:$startLabel {id: $startId})-[:$relType*2]->(target)
RETURN target

Example - Friends of Friends:

cypher

MATCH (alice:Person {name:"Alice"})-[:FOLLOWS*2]->(fof:Person)
RETURN fof

Use Cases:

• Exactly 2 hops from Alice
• Precise distance queries
• Specific relationship chain discovery

---

2. Variable Depth Multi-Hop

Pattern: Traverse between MIN and MAX relationships

Cypher:

cypher

MATCH (start:$startLabel {id: $startId})-[:$relType*1..$maxHops]->(target)
RETURN target
LIMIT $limit

Example - Up to 3 Hops:

cypher

MATCH (company:Company {id:"C123"})-[:SUPPLIES*1..3]->(supplier)
RETURN supplier
LIMIT 50

Use Cases:

• Within N hops
• Range-based discovery
• Flexible connection finding

---

3. Path Discovery

Pattern: Return actual path structures

Cypher:

cypher

MATCH path = (start:$startLabel)-[*1..$maxHops]-(target:$targetLabel)
RETURN path, LENGTH(path) as hops
LIMIT $limit

Example - Path from Alice to Bob:

cypher

MATCH path = (alice:Person {name:"Alice"})-[*1..4]-(bob:Person {name:"Bob"})
RETURN path, LENGTH(path) as hops
LIMIT 10

Returns: Full traversal paths showing relationships

---

4. Filtered Multi-Hop

Pattern: Multi-hop with relationship type filtering

Cypher:

cypher

MATCH (start:$startLabel)-[:$relType*1..$hops]->(target:$targetLabel)
WHERE start.id = $startId AND target.$property = $value
RETURN target
LIMIT $limit

Example - Specific Relationship Chain:

cypher

MATCH (source:Person)-[:KNOWS|:WORKS_WITH*1..3]->(target:Person)
WHERE source.name = "Alice" AND target.active = true
RETURN target
LIMIT 25

---

5. Undirected Multi-Hop

Pattern: Follow relationships in any direction

Cypher:

cypher

MATCH (start:$startLabel)-[:$relType*1..$maxHops]-(target)
RETURN target

Example - Undirected Network:

cypher

MATCH (node:Entity)-[*1..3]-(connected)
RETURN connected
LIMIT 100

---

6. SPARQL Property Paths

Pattern: SPARQL+ operator for one or more hops

SPARQL:

sparql

SELECT ?target
WHERE {
  ?start ex:$property+ ?target .
}

Example - Property Path:

sparql

PREFIX ex: <http://example.org/>
SELECT ?person
WHERE {
  :Alice ex:knows+ ?person .
}

---

7. Aggregated Multi-Hop

Pattern: Multi-hop with aggregation

Cypher:

cypher

MATCH (start:$startLabel)-[:$relType*1..$hops]->(target)
RETURN target, COUNT(*) as path_count
GROUP BY target
ORDER BY path_count DESC

Example - Count Paths:

cypher

MATCH (alice:Person {name:"Alice"})-[:FOLLOWS*1..3]->(person)
RETURN person, COUNT(*) as num_paths
GROUP BY person
ORDER BY num_paths DESC
LIMIT 10

---

8. Conditional Multi-Hop

Pattern: Multi-hop with WHERE conditions

Cypher:

cypher

MATCH path = (start:$startLabel)-[:$relType*1..$hops]->(target)
WHERE ALL(n in nodes(path) WHERE n.$property <> $excludeValue)
RETURN target

Example - Avoid Specific Nodes:

cypher

MATCH path = (alice:Person {name:"Alice"})-[:KNOWS*1..3]->(person)
WHERE ALL(n in nodes(path) WHERE n.status = "active")
RETURN person
LIMIT 50

---

📊 Multi-Hop Complexity

Hop Depth Impact:

text

Depth 1: O(n)
  ├─ 1 hop from start node
  └─ Direct relationships only

Depth 2: O(n²)
  ├─ 2 hops from start node
  └─ Friends of friends

Depth 3: O(n³)
  ├─ 3 hops from start node
  └─ Can be expensive

Depth 4+: O(n^4+) ⚠️ EXPENSIVE
  ├─ Exponential growth
  └─ Use only with filtering/limits

Performance Guidelines:

• Keep depth ≤ 3 for open queries
• Use depth ≤ 4 with strong filters
• Avoid depth > 4 without restrictions
• Always include LIMIT clause

---

🏢 Real-World Examples

Social Network: Friends of Friends

cypher

MATCH (user:User {id: $userId})-[:FOLLOWS*2]->(fof:User)
WHERE NOT (user)-[:FOLLOWS]->(fof)
RETURN DISTINCT fof
LIMIT 50

E-Commerce: Product Recommendations

cypher

MATCH (customer:Customer {id: $customerId})-[:PURCHASED*1..2]->(recommended:Product)
RETURN recommended, COUNT(*) as relevance
GROUP BY recommended
ORDER BY relevance DESC
LIMIT 10

Supply Chain: Supplier Network

cypher

MATCH (company:Company {name: $companyName})-[:SUPPLIES*1..3]->(supplier:Company)
RETURN supplier, COUNT(*) as connections
GROUP BY supplier
ORDER BY connections DESC
LIMIT 20

---

✅ Best Practices

When building multi-hop queries:

1. Limit Traversal Depth

   • Depth ≤ 3 for unrestricted queries
   • Depth ≤ 4 with heavy filtering
   • Always bound depth

2. Restrict Relationship Types

   • Specify exact relationships
   • Avoid wildcard traversal
   • Filter by direction

3. Use Result Limits

   • Always include LIMIT clause
   • Prevents memory issues
   • Controls response size

4. Start from Specific Nodes

   • Use indexed properties for entry
   • Filter by specific IDs/properties
   • Avoid full scans

5. Filter Within Paths

   • Use WHERE with ALL/ANY
   • Exclude specific nodes
   • Filter by node properties

6. Apply Indexes

   • Index starting node properties
   • Index relationship types
   • Index filter properties

7. Test Performance

   • Profile with real data
   • Monitor execution time
   • Optimize entry points

8. Document Assumptions

   • Expected path lengths
   • Relationship cardinality
   • Performance characteristics

---

🔗 Integration with Other Skills

This skill integrates with:

• graph-query-debugging-tool - Debug multi-hop queries
• graph-query-optimization-assistant - Optimize multi-hop patterns
• graph-template-query-generator - Generate multi-hop templates
• nl-to-graph-query-translator - Translate natural language to multi-hop
• graph-schema-validation - Validate multi-hop against schema

---

📋 Output Formats

The skill returns:

1. Multi-Hop Query

Executable Cypher or SPARQL query with specified hop depth.

2. Path Query

Returns relationship paths showing traversal chains.

3. Parameterized Template

Reusable template with hop parameters.

4. Complexity Analysis

Performance metrics and recommendations.

---

💡 Reasoning Patterns

Pattern 1: Find Connections

Discover all entities within N hops.

cypher

MATCH (start)-[:RELATIONSHIP*1..$n]->(end)
RETURN end

Pattern 2: Find Paths

Return actual paths showing relationship chains.

cypher

MATCH path = (start)-[*1..$n]-(end)
RETURN path

Pattern 3: Count Paths

Count number of different paths.

cypher

MATCH (start)-[:REL*1..$n]->(end)
RETURN end, COUNT(*) as paths

Pattern 4: Shortest Path

Find shortest connection.

cypher

MATCH path = shortestPath((start)-[*]-(end))
RETURN path

Pattern 5: Circular Detection

Find cycles in relationships.

cypher

MATCH path = (start)-[:REL*2..]->(start)
RETURN path

---

📖 Summary

This skill generates multi-hop graph queries that traverse multiple relationships to uncover indirect connections and hidden patterns in graph datasets.

It enables developers to perform advanced graph reasoning using Cypher or SPARQL for social networks, recommendation systems, fraud detection, and knowledge graph exploration.

---

Status: Enterprise-Grade Multi-Hop Reasoning

Comprehensive multi-hop query generation and reasoning system for professional knowledge graph development.