While Knowledge Graphs (KGs) have been widely adopted in the building industry, graph representations do not display topological relationships in an intuitive manner. Furthermore, advanced Building Information Management (BIM) models are often not available, meaning this KG construction is often done manually. Even if a BIM model is available, this model contains all information, meaning the KG is unnecessarily large for most tasks.
The typical graph representation of a floor of an office building does not intuitively display the topological relationships of the rooms, making it hard to know whether everything modelled in the knowledge graph is correct.
The goal of this thesis is to create a user interface that allows for intuitive display of buildings modelled as KGs. Ideally, one could either type new triples that get visualized on the plan, or draw something that gets added as a triple. This idea is very similar to the Mermaid UI, which allows drawing charts using markdown.
When interesting, data for the KG can also be mined from related building documents, such as Energy Performance Certificates (EPCs) or maintenance certificates of technical installations.
To achieve this, the student could research the following steps:
-
Derive both functional and non-functional requirements for the tool
-
Research building-related ontologies (such as BOT, BEO, SAREF4BLDG, etc.) and their alignment
-
Implement a prototype that allows for more intuitive KG construction for buildings, optionally allowing to extract metadata from incoming building-related documents (EPCs, maintenance certificates, etc.)
-
Define and conduct an evaluation of the developed prototype (performance, usability improvement compared to manual graph creation)