A deep learning approach to Question Answering systems for the digital twins of buildings

The data describing different aspects of a building comes from various disciplines and is most often residing in disparate systems. Yet, an effective management of the building performance requires access to a unified view of the whole lifecycle data. Using existing technologies disparate sources of the building data can be integrated and stored in the form of graphs whose nodes represent building entities and whose edges describe the relationships between the entities. However, retrieving data from such graphs requires technical knowledge about database systems and the backbone of the graphs (i.e., data schemas). Hence, applicability of such technologies remains inefficient for building professionals. This research proposes to investigate a highly intuitive interface which allows users to search through the connected sources of building data using natural language statements. The results can lay a foundation for developing commercialized products of intelligent conversational agents for industrial applications.

Faculty Supervisor:

Ali Motamedi

Student:

Partner:

Beslogic

Discipline:

Engineering

Sector:

Professional, scientific and technical services

University:

École de technologie supérieure

Program: