Colibri provides a platform for smart building energy management. Semantics about the building, the building automation systems, other energy-consuming or energy-producing devices, and the environment is used to elaborate optimization strategies. Information is exchanged with the building automation systems, which influence physical processes in the building, and with external agents, such as smart grid agents or Web service providers.

In this project, a hybrid data store should be implemented in order to separate the (static) semantics and the exchanged (runtime) data for performance reasons. While semantically enriched information will be stored in a native triple store, the runtime information (e.g. sensed temperature values, received messages from smart grid agents) is rolled out in a relational database. Apache Jena is used as triple store, and PostgreSQL is used as relational database. Access to this hybrid data store should be designed as transparent as possible imitating a mere triple store. Thus, querying and modification of information is based on Semantic Web technologies instead of SQL commands. The developed data store management system needs to be able to interpret these queries and combine information from both internal data stores.

Benefit for the Student

Dive into latest technologies and emerging protocols for the Semantic Web. Improve your experiences in Java-based software development and database systems.

Benefit for the Project

The potentially huge amount of information that needs to be processed in order to provide energy-efficient operation of buildings exceeds the capability of classical triple stores. A hybrid approach consisting of an ontology for storing semantically enriched information and a classical, relational database for plain data will overcome this issue. Colibri optimization will benefit from the performance gain.


Strong skills in Java-based software development are necessary. In addition, skills in Semantic Web technologies (SPARQL, OWL) and relational databases are required.


Daniel Schachinger, Andreas Fernbach, Wolfgang Kastner

More information (Colibri) (SPARQL) (PostgreSQL) (Apache Jena) (OWL)