Business context
The client builds software for telecom network operators. Network assets such as access points, poles, towers and optical fiber cables sit on and inside real estate: sites, buildings, floors and rooms. Their product needed to connect the two to answer planning and operations questions.
The challenge
Network data and real estate data lived in separate systems, with different IDs and structures. Relationships such as which access point serves which space, or which cable runs along which poles, were implicit or missing. A fixed relational schema could neither capture these links flexibly nor infer new ones.
Our solution
We built a knowledge graph proof of concept on AWS Neptune. The RealEstateCore ontology models buildings, spaces and equipment, and we extended it with telecom network concepts. Neptune stores the ontology and the data instances together as RDF, queried with SPARQL. Reasoning over the graph derives relationships that are not stated explicitly, such as network dependencies and coverage.
- Ontology based on RealEstateCore, extended with access points, poles, towers and optical fiber cables
- Ontology and data instances captured in AWS Neptune as a graph database
- Reasoning to infer dependency (what a fiber route or asset depends on) and coverage (which spaces each asset serves)
- Open, standards-based model rather than a proprietary schema
Business impact
- Validated the knowledge graph approach on real telecom and real estate data before the client committed to a production build
- One connected model for questions that span network and property data, such as which assets serve a building or what a fiber route depends on
- A standard ontology eases integration with building and smart city systems that also use RealEstateCore
- A graph foundation that AI agents can query for connected, scoped context
Engagement
A multi-month proof of concept.
