More than numbers: Using data to help industries “go digital”

Amidst challenging economic times and digitalisation, the business community and numerous organisations are using digital technologies to reinvent operations, enabling them to compete and thrive in markets and industries. Digital transformation comes with a practical purpose: to assist organisations in automating processes, delivering customisable services and products, and responding agilely to changes. Before these can be achieved, however, we must first delve into organisational data, study patterns, identify areas for improvement, and then develop computing algorithms for automation, customisation and agility enhancement.

Making sense of complex data

Every day, we are bombarded with data from various channels. This data comes in different sizes, forms (e.g., texts, numbers, videos, images) and speeds (e.g., real-time, time lag). The quality can vary, as some sets of data are less reliable than others. Data science is about making sense of diverse data. Prof. Li studies Multi-modal Data Management and Data Warehousing & Mining in particular.

“Multi-modal Data Management refers to the processing and use of multiple types and sources of data with a single information management system that can integrate, transform, index, store and analyse these data,” Prof. Li said. “Data Warehousing is about uploading, cleansing and organising various operational and transactional data into one repository. This involves building appropriate data marts so that data can be derived and transformed in a way that is easily understandable and manipulatable by end-users, who may be middle-level managers and/or senior decision-makers. Data Mining is fetching important data from these databases or warehouses and finding rules and patterns hidden in the data.”

Data management is a broad concept that concerns the flow of information across different stages in its life cycle, from data creation, storage, usage, sharing, and archiving to destruction. Data management is closely related to an organisation’s technology infrastructure, quality control, and governance. Data needs to be managed safely and ethically so that it can be transformed into insights and knowledge for driving organisational growth.

Data science as a powerful, value-adding discipline

Data is the foundation on which industries craft new solutions that help them compete and succeed. The manufacturing industry, for example, uses analytics to predict potential equipment defects, so that timely repairs can be made before machine failures occur. In healthcare, clinics and hospitals leverage electronic health records of patients to improve diagnoses and identify flu trends. Health problems can thus be addressed proactively before they escalate and snowball into a public health crisis. In the retail market, shop owners combine information about transactions and product inventories to better understand sales and tailor their products to customer tastes and choices.

“Data is a strategic asset for any industry and enterprise,” Prof. Li said, “the value of data is not just based on its sources, quality or formats, but also very much depends on how and what we do with it. We seek to deliver value through a greater understanding, alignment and actioning of data”. Data science is an interdisciplinary field, involving different bodies of knowledge (e.g., scientific methods, processes, algorithms, and systems) and techniques (e.g., statistics, machine learning, data visualisation, and programming) for analysing a large data set.

Revolutionising the travel industry with data-led transformation

In January 2024, PolyU established the Research Centre for Digital Transformation of Tourism (RCdTT) as a new constituent unit under PAIR. Prof. Li, appointed as RCdTT Co-Director, is determined to steer the Centre toward new data-led tools and technologies that are conducive to the economic and sustainable development of tourism.

The establishment of RCdTT is a response to the call for “optimised” travel experience and ESG (Environment, Social and Governance). “The aim of RCdTT is to reshape tourism and hospitality operations, destination governance, and international collaboration to promote sustainable travel experiences and business practices,” Prof. Li explained. RCdTT combines PolyU expertise across disciplines, including hotel and tourism management, engineering, applied mathematics, land surveying and geo-informatics, design, computing, management and marketing, for data-led interdisciplinary solutions that help the travel industry “go digital”.

“To do so, the Centre is embarking on research with two foci: a Digital Tourism Decision-making and Monitoring System, and AI-Driven Business and Experience Innovations,” Prof Li said. “The former focuses on the development of digital systems and tools for monitoring carbon emissions, tourist satisfaction and sustainability in the industry. The latter focuses on innovating tourism and consumer experiences using artificial intelligence, robotics and immersive reality.”

Beyond tourism: Transcending impact of data science

At PAIR, Prof. Li contributes his computing expertise extensively. “With the fundamental importance of artificial intelligence and data science, I have been involved in several PAIR units,” he shared. Currently, Prof. Li is a Management Committee Member of the Research Institute for Artificial Intelligence of Things (RIAIoT) and the Otto Poon Charitable Foundation Smart Cities Research Institute (SCRI), an International Advisory Committee Member of the Research Institute for Advanced Manufacturing (RIAM), and a Member of the Research Institute for Sports Science and Technology (RISports).

“Definitely, my engagements with PAIR have widened my exposure to multi-disciplinary fields. It is amazing to see that almost every domain and field nowadays has to deal with data in different ways,” Prof. Li said. He sees AI algorithms, when coupled with big data techniques, as tools for powerful solutions and ubiquitous intelligence.

Data science for “better” smart vehicles

Vehicular edge computing (VEC) is an emerging technology in the automotive industry. It allows data coming from devices connected to the vehicles to be processed in a faster way so that “immediate” and “near-real time” information can be provided to drivers. The data cache, i.e., temporary storage of files that normally helps reduce retrieval time when an application is next used, results in additional energy consumption. VEC is facing a trade-off between performance and energy consumption.

At RIAIoT, Prof. Li focuses on two specific research areas, AIoT Analytics and Cross-layer Issues. Together with PolyU collaborators and other overseas collaborators in China and Portugal, he formulated a new mathematical model for cache-assisted VEC, which has demonstrated its ability to dramatically optimise the response time, while keeping the energy consumption across time at a low level. The study, “Toward Response Time Minimization Considering Energy Consumption in Caching-Assisted Vehicular Edge Computing”, was published in the IEEE Internet of Things Journal.

Driving ESG in logistics

Environmental, Social and Governance (ESG) is a buzz-phrase in the business world, but a data system to reliably measure ESG performance has been lacking. Currently, the logistics industry is faced with increasing demands and expectations from government and investors for fulfilling ESG. However, there is a need for more transparent reporting of the industry’s ESG performance.

Prof. Li and his RIAIoT colleagues are determined to address this need through a large project titled “Environmental, Social and Governance (ESG) Management – A System Design Approach”. The project aims to design and develop an ESG management system using emerging technologies. The system will be used in and applied to the Hong Kong logistics industry.

The concept of ESG is pure and simple, but the measurement of it can be very complex. How should we define “sustainability”? What kind of data needs to be collected? How can we ensure the trustworthiness of the data? What kinds of technologies can support real-time data collection? To develop the system, the team must first outline the meta-requirements and then identify specific design and enabling technologies that fulfil these requirements. The innovation, a smart logistics management system platform for ESG reporting, will combine blockchain, internet of things and artificial intelligence, thus enabling automated ESG data collection and digitalisation.

Behind rationality: Data scientist with a love for young people and education

It is generally believed that data scientists must think so logically and analytically about rules and systems that they exclude emotions from their scientific endeavours. However, Prof. Li utilises his mathematical logic in caring about others. He has successfully developed a data-based method for detecting suicidal ideation among young people based on fusing and mining of multi-modal data over the internet and social media. The data set is going to be opened for public access, so that other institutions can make use of the resource to stop more tragedies from happening.

In addition, Prof. Li and his colleagues in the Department of Computing have been working on a virtual teaching assistant called “Data Buddy”. “For academics, effective and good-quality education is naturally our basic duty and responsibility,” he said. “University education is no longer just about giving lectures and assignments to teach ‘static knowledge’ that can be acquired easily online through websites and AI tools nowadays. ‘Teaching how to teach’ and ‘learning how to learn’ will become more important than ever.”

Data Buddy is an AI-powered technology that structures and visualises all the learning content of a particular university subject on a Knowledge Graph, acquires information about the student users’ learning and career goals through a Q&A function, and guides the users to design their own study plans. Prof. Li’s team is now working on the next phase of development, which involves capturing behavioural data of students and teachers, so that Data Buddy can be upgraded with more powerful, engaging functions.

Knowing your limit when keeping up with changes in computer science

Prof. Li had worked at another local university for 20 years before re-joining PolyU in 2018 (he was previously a junior academic of PolyU, from 1997 to 1998). Currently, he is the Chairperson of the Hong Kong Web Society and holds important roles in various computing associations. He has served as a consultant to enterprises and institutions worldwide, including Microsoft and Motorola.

Computer science is a fast-changing, competitive field. Scientists need to spot and keep abreast of emerging technologies, and not get lost among the many new things around them. “For young researchers who want to be successful, it is crucial that they concentrate on a couple of topics and stick with them. Avoid topics that are too mature and have already been studied by many people. Likewise, avoid brand-new topics that are too ‘hot’ and place a high demand on hardware, because the industry has a competitive advantage over academic scholars,” Prof. Li advised. “Interdisciplinary collaboration is useful, but a young researcher should put up a fence to avoid expending too much time and energy.”

Over the decades, Prof. Li has witnessed several IT revolutions, from the focus on the processing of simple, structured data, to the heuristic modelling of complex data as the internet emerged, to the present-day automated, simultaneous processing of “big data” coming in various modalities. As technology evolves, so do our expectations of human skills. “Knowing how to embrace this fast development and keep up the pace will become an essential skillset,” Prof. Li explained.

While things keep changing, there is one thing that does not: the fundamental significance of data. According to Prof. Li, “Data science still has a long way towards the peak of its ‘golden era’.”