Aviation Safety Use Case¶
Analysis of aviation safety incidents, accidents, and related data using knowledge graphs for pattern detection and investigation.
Overview¶
| Aspect | Value |
|---|---|
| Domain | Aviation incident investigation |
| Data Format | Structured Markdown incident reports |
| Ontology | ECCAIRS-derived aviation taxonomy |
| Incidents | 10 European aviation occurrences (2024) |
| Relationships | Aircraft, operators, airports, causes |
Purpose¶
This use case demonstrates Aletheia's capabilities for:
- Semantic search over narratives: Finding incidents by natural language descriptions
- Multi-hop relationship queries: Connecting aircraft → operators → incidents → causes
- Cause-and-effect analysis: Understanding primary causes and contributing factors
- Geographic and temporal analysis: Filtering by location and time
Data Coverage¶
Airports¶
| Airport | ICAO | Incidents |
|---|---|---|
| Paris Charles de Gaulle | LFPG | 1 |
| Frankfurt | EDDF | 1 |
| Nice Cote d'Azur | LFMN | 1 |
| Lyon Saint-Exupery | LFLL | 1 |
| Madrid Barajas | LEMD | 1 |
| Milan Malpensa | LIMC | 1 |
| Palma de Mallorca | LEPA | 1 |
| Amsterdam Schiphol | EHAM | 1 |
| Barcelona El Prat | LEBL | 2 |
Airlines¶
- Air France, HOP! Air France
- Lufthansa
- KLM Cityhopper
- TAP Air Portugal
- ITA Airways
- Iberia
- Vueling
- Air Europa
Aircraft Types¶
- Airbus: A320-214, A321neo, A330-200
- Boeing: 737-800, 787-9
- Regional: CRJ-1000, ATR 72-600, ERJ-195, Fokker 70
Incident Types¶
| Type | Example |
|---|---|
| System failure | Hydraulic pump failure |
| Weather | Clear air turbulence, wind shear |
| Wildlife | Bird strike |
| Software | Display management anomaly |
| Maintenance | FOD damage from ground equipment |
| Human factors | Smoke in cockpit (MAYDAY) |
Key Capabilities¶
1. Semantic Cause Lookup¶
Question: "What caused the turbulence-related injuries near Barcelona?"
Answer: Unpredicted clear air turbulence associated with jetstream boundary
2. Location-based Discovery¶
Question: "What incident occurred at Nice Cote d'Azur Airport?"
Answer: Wind shear encounter during approach caused by catabatic winds
3. Entity Description¶
Question: "Describe incident 2024-0412-EU"
Answer: A TAP Air Portugal Embraer ERJ-195 encountered severe clear air
turbulence near Barcelona, injuring two cabin crew members
4. Multi-hop Queries¶
Question: "What incident involved the ITA Airways ATR 72?"
Answer: Display management computer software anomaly during approach
to Milan Malpensa
ECCAIRS Taxonomy Support¶
This use case supports the ECCAIRS (European Co-ordination Centre for Accident and Incident Reporting Systems) aviation taxonomy:
- Standardized terminology for occurrence reporting
- Used by EASA and national authorities
- Convertible to OWL ontology format
Why This Dataset?¶
Aviation incident data is ideal for GraphRAG because:
- Rich narratives: Detailed incident descriptions for semantic search
- Complex relationships: Multiple entities connected per incident
- Domain-specific terminology: Tests technical vocabulary handling
- No parametric knowledge: Specific incidents not in LLM training
- Cause-effect chains: Primary cause → contributing factors → outcomes
Documentation¶
| Page | Description |
|---|---|
| Functional Guide | How to use this use case |
| Technical Reference | Implementation details |
| Demo Script | Step-by-step demonstration |
Quick Start¶
# 1. Load ontology (optional but recommended)
aletheia build-ontology-graph \
--use-case aviation_safety \
--knowledge-graph aviation_safety_ontology
# 2. Build knowledge graph
aletheia build-knowledge-graph \
--use-case aviation_safety \
--knowledge-graph aviation_safety \
--schema-mode graph-hybrid \
--ontology-graph aviation_safety_ontology
# 3. Run evaluation
aletheia evaluate-ragas \
--knowledge-graph aviation_safety \
--questions use_cases/aviation_safety/evaluation_questions_curated.json \
--grounding-mode strict