Chuyển tới nội dung chính

Neo4j Query Agent

The Neo4j Query Agent provides high-level graph query operations for traffic analytics and relationship analysis.

📋 Overview​

PropertyValue
Modulesrc.agents.graph_database.neo4j_query_agent
ClassNeo4jQueryAgent
AuthorUIP Team
Version1.0.0

🎯 Purpose​

The Neo4j Query Agent enables:

  • Complex graph queries using Cypher language
  • Path finding between traffic entities
  • Aggregation queries for analytics dashboards
  • Real-time graph traversal for pattern detection

🔧 Query Types​

Entity Queries​

# Get camera with relationships
agent.get_camera_with_observations("CAM_001", limit=100)

# Get accident with related entities
agent.get_accident_context("ACC_001")

Path Queries​

# Find shortest path between cameras
agent.find_path("CAM_001", "CAM_050", max_hops=5)

# Find all paths through congestion zones
agent.find_congestion_paths(start_camera, end_camera)

Analytics Queries​

# Get hotspot cameras
agent.get_accident_hotspots(min_accidents=5)

# Get traffic flow patterns
agent.get_flow_patterns(time_range="24h")

📊 Query Examples​

Find Nearby Cameras​

from src.agents.graph_database.neo4j_query_agent import Neo4jQueryAgent

query_agent = Neo4jQueryAgent()

# Find cameras within spatial proximity
nearby = query_agent.find_nearby_cameras(
    camera_id="CAM_001",
    max_distance_km=1.0
)

Traffic Pattern Analysis​

# Analyze morning rush patterns
patterns = query_agent.analyze_patterns(
    time_window="07:00-09:00",
    days=["Monday", "Tuesday", "Wednesday"]
)

# Get congestion correlations
correlations = query_agent.get_congestion_correlations(
    threshold=0.7
)

Accident Impact Analysis​

# Find all entities affected by accident
impact = query_agent.get_accident_impact("ACC_001")

# Result includes:
# - Affected cameras
# - Traffic flow changes
# - Congestion ripple effects
# - Estimated recovery time

🚀 Usage​

Initialize Agent​

from src.agents.graph_database.neo4j_query_agent import Neo4jQueryAgent

config = {
    "enabled": True,
    "neo4j_uri": "bolt://localhost:7687",
    "username": "neo4j",
    "password": "password"
}
query_agent = Neo4jQueryAgent(config)

Execute Custom Queries​

# Run custom Cypher query
result = query_agent.execute_query("""
    MATCH (c:Camera)-[:OBSERVES]->(o:Observation)
    WHERE o.timestamp > datetime() - duration('PT1H')
    RETURN c.name, avg(o.vehicle_count) as avg_count
    ORDER BY avg_count DESC
""")

# Run with parameters
result = query_agent.execute_query(
    query="MATCH (c:Camera {id: $id}) RETURN c",
    params={"id": "CAM_001"}
)

Aggregate Functions​

# Get statistics
stats = query_agent.get_statistics()
# Returns:
# {
#     "total_cameras": 150,
#     "total_observations": 45000,
#     "total_accidents": 234,
#     "avg_observations_per_camera": 300
# }

⚙️ Configuration​

# config/neo4j_sync.yaml
neo4j_query:
  enabled: true
  
  # Query optimization
  optimization:
    use_indexes: true
    cache_results: true
    cache_ttl: 60
  
  # Query limits
  limits:
    max_results: 10000
    timeout_seconds: 30
    max_path_length: 10

📈 Performance Optimization​

Index Usage​

// Ensure indexes exist
CREATE INDEX camera_id_index FOR (c:Camera) ON (c.id)
CREATE INDEX observation_timestamp_index FOR (o:Observation) ON (o.timestamp)

Query Profiling​

# Profile query performance
profile = query_agent.profile_query("""
    MATCH (c:Camera)-[:DETECTS]->(a:Accident)
    RETURN c, a
""")
# Returns execution plan and timing

📖 Related Documentation​

See the complete agents reference for all available agents.