Research

"Nothing has such power to broaden the mind as the ability to investigate systematically and truly all that comes under thy observation in life."

- Marcus Aurelius -

Current research projects

MEng Project

2021 - 2022

Web-based application for real-time progress data visualisation with BIM 4D in IBS construction
Student: Tan Kwan Yee
Supervisor: Dr Jing Ying Wong

The challenge
Centralized real-time tracking of IBS-related construction progress is a daunting task for stakeholders on a construction project. 

Proposed Solution
The project utilises JavaScript architectures that are readily available to construct an app that tracks the progress of IBS construction progress in real-time. The app allows progress monitoring through data visualisations in the form of 4D BIM and time charts. Also, the app has a simple site diary entry interface for capturing other semantics other than IBS construction progress.

PhD Project

2020 - Date

An Adaptive Data Processor for the De-silo-fication of Mobility-as-a-Service
Data

Student: Shams Ghazy
Main Supervisor: Dr. Jing Ying Wong
Co-Supervisor: Mr. Yu Hoe Tang
Co-Supervisor: Prof. Andy Chan
External Co-Supervisor: Dr. Pieter Colpaert

 
The challenge
In the transportation domain, each mobility service generates data. The data is commonly stored in a certain
structure, with inherent semantics that describe the data, and different access rules. Mobility-as-a-Service (MaaS) aggregates data from heterogeneous sources, not only transport data of varying types of mobility, but also factors affecting the success of a user’s trip.

 
Every time a MaaS provider integrates a transport operator, they are expected to merge their data into the MaaS ecosystem. This process is not standardized as operators commonly follow different standards better suited for their own services and business rules. Consequently, the MaaS provider needs to configure their services to talk to several custom APIs, resulting in an unscalable and inefficient process, hindering the progress of MaaS.
 
Currently, there exists a few attempts in standardizing this process such as the TOMP-API, however, the process remains highly dependent on the bilateral agreement a MaaS provider forms with an operator, and hence lacks modularity and adaptability for reuse with the next operator. In addition, there is no clear method for an equitable sharing of user data, precipitating a concern for the operators to take part in a MaaS ecosystem in the first place. 
 
Proposed solution
This research proposes the use of Semantic Web Technologies as a solution to formulate an adaptive and interoperable data exchange process between MaaS stakeholders taking into account their heterogeneous interoperability requirements.

PhD Project

2018 - Date

A framework for Knowledge Representation Learning on graph-based building control data
Student: Kevin Luwemba Mugumya
Main supervisor: Dr Jing Ying Wong
Co-supervisor: Prof Andy Chan 

The challenge
As soon as a building is commissioned, a chain of events is set into motion to ensure proper functionality of its systems and that operational efficiency targets are met in compliance with set regulations. This process is called Facility Management (FM) and just like any other stage of a building’s life-cycle, FM is a heavily data-driven process that involves several building systems and multi-disciplinary stakeholders constantly exchanging disintegrated and scattered datasets.

 

Dealing with siloed data is challenging and to intuitively visualize this, consider the following questions in a building's context.

  1. How much of the data and information used to manage a building is dispersed across its different systems, zones, components, hard copy documentation or databases?

  2. How many clicks and reports do facility managers have to navigate in the interface of a Building Automation System to find an answer to a single FM problem or get relevant results for a specific search?

  3. How efficient is it to optimize a building's FM process using Machine Learning (ML) models that are trained using disintegrated datasets? 

Proposed Solution

This research leverages Semantic Web Technologies (SWT) to orchestrate a mechanism for unifying heterogenous FM datasets with enough expressivity for intuitive querying and inferencing (logical and machine learning-driven) in downstream FM tasks specifically building automation. This research also investigates how to leverage the inherent relational structure of semantically inter-linked FM datasets as a mechanism for message passing and information propagation to facilitate collective contextual reasoning* in building automation agents.

* Inter-linked data exhibits patterns and dependencies that occur between attributes and relationships of different entities of the dataset. ML methods that can exploit these patterns collectively in their learning pipeline are referred to in this research as exhibitants of collective contextual reasoning.