Learning at Zhejiang University: Applying the Learning Engineering Approach to ICT-Enabled Educational Innovation

Abstract

Learning engineering is an interdisciplinary field that uses learning sciences, specifically human-centered engineering design and data-informed decision-making, to support learners and their development. Its theoretical foundation is a blend of scientific disciplines, including learning sciences, data science, computer science, and instructional system design, and it uses a holistic engineering methodology. A learning engineering approach considers the full learning cycle, linking together divergent modules to achieve scalable solutions. As a “team sport” that requires multidisciplinary expertise and integrated efforts across different sectors, it functions as an inclusive ecosystem. This paper introduces the successful implementation of Learning@ZJU, a novel educational system based on a learning engineering approach, at Zhejiang University in China. The development included proposing a new theoretical framework (K-CPS, named for Knowledge graph, smart Classrooms, learning Platforms, and cloud Services), establishing a full-spectrum, cross-sector education system, and building a collaborative ecosystem that integrates education, research, and industry. Although the setting of this study is China, which has the largest student enrollment in higher education in the world, the system has the potential to influence the global education model.

Introduction

What is Learning Engineering?

Carnegie Mellon University (CMU) is the home of learning engineering. The term “learning engineer” was first proposed in the 1960s by the late Herbert Simon, a Nobel and Turing Award laureate and CMU professor:

There is no simple path that will take us immediately from the contemporary amateurism of the college to the professional design of learning environments and learning experiences. There are, however, some obvious first steps along the way. The most important step is to find a place on the campus for a team of individuals who are professionals in the design of learning environments—learning engineers, if you will. (Simon, 1967)

Evolving from this embryo, learning engineering has more recently been described as one of the “applied learning sciences” (Kessler et al., 2023). Victor R. Lee (2023) cited Janet Kolodner's definition of learning engineering:

Learning engineering is the systematic application of principles and methods from the learning sciences to support and improve our understanding of learners and learning process…Learning engineers design learning experiences and environments informed by the learning sciences. They combine knowledge, tools, and techniques from a variety of technical, pedagogical, empirical, and design-based disciplines while collaborating with subject-matter experts, software engineers, and others.

More recently, the Industry Connections (IC) program of the Institute for Electrical and Electronics Engineers Standards Association (IEEE-SA) founded the IC Industry Consortium on Learning Engineering (ICICLE) professional organization. ICICLE describes learning engineering as a process and interdisciplinary practice that uses human-centered engineering design methodologies and data-informed decision-making to support learners and their development (Retrieved from ICICLE—Learning Engineering (ieee.org)).

The LearnLab at CMU describes learning engineering as a way to design and build learning environments that make in-person, online, and hybrid classes more effective. Learning engineers apply science of learning principles, evidence-based research, qualitative and quantitative cognitive task analysis, and data-driven methods to design, create, and improve educational resources and technologies that enable students and instructors to succeed (Retrieved from https://learnlab.org/introduction-to-learning-engineering/) (Figure 1).

Figure 1.

Learning Engineering Cycle (Retrieved from https://learnlab.org/introduction-to-learning-engineering/).

Simon Initiative

Carnegie Mellon University faculty and teaching support professionals have been at the forefront of learning science for decades, developing evidence-based educational technologies and taking pains to apply the principles of learning engineering to their teaching practice (Smith & Herckis, 2018).

Responding to Herbert Simon's call for a systematic and scientific approach to teaching and learning, CMU launched the Simon Initiative to nurture a cross-disciplinary learning engineering ecosystem. The Initiative's goal, developed over several decades, has been to measurably improve student learning outcomes. Its work constitutes a continuous cycle in which learning science informs educational practice and data from educational practice inform learning science, leading to better learning outcomes for students from all backgrounds in any place.

Information and Communication Technology (ICT) in Education

Responses to ICT in Education

Research on the impact of technology in education has burgeoned since the advent of ICTs in the 1960s. The literature on the development and use of educational technologies has expanded exponentially over the past decades and has included the emergence of specialized journals, such as Educational Technology Research and Development, Educational Technology Review, and The International Journal of Educational Technology, and specialized conferences, such as the Association for Educational Communications and Technology Convention, Artificial Intelligence in Education, and the International Conference on Intelligent Tutoring Systems.

Data on the impact of technology on learning and teaching are demanded by those who determine what funding will be available to pay for technology within education systems (Cukurova & Luckin, 2019). Due to rapid technological changes, national strategies and policies for higher education require ongoing research on how ICT supports learning and research (Conole, 2002). This provides immense opportunities, but makes firm commitments to specific systems or developments very difficult. Government policymakers in the education field must keep their eyes on the evolving technological environment and try to proactively direct the implementation of new technologies in education.

China, which has the largest number of students in higher education in the world (see the white paper “Chinese Youth in the New Era” issued by the State Council Information Office of China in 2022), is no exception to these trends. In response to the massive impact of technology on education and the extensive research on educational technology, the Chinese government has, since 2018, issued a series of strategic documents such as China's Education Modernization 2035, the ICT-based Education Action Plan 2.0, and the Guidelines on Promoting the Construction of New Education Infrastructure and Building a High-quality Education Support System. The government also launched its National Strategy Action Plan for Digitization of Education to accelerate the digital transformation of education and support the development of high-quality education through the use of ICT in education.

How does technology impact education? How is research applied in educational settings? How are the relevant national policies implemented? These questions direct our attention to the education context where these changes are happening, that is, institutions of higher education. Research has shown that over the past decade, educational technologies have been integrated into the core business of institutions and are now part of a wide, generic (i.e., not subject-specific) debate on learning and teaching. As no single article could adequately cover the breadth and scope of this debate, this paper does not offer a panoramic overview of the actions taken by institutions or a systematic review of the literature. Instead, it scrutinizes the status quo of ICT-based education in China and introduces a modest but self-justified system that reflects current best practices.

Challenges of ICT in Education

China's higher education practices undoubtedly provide insights that can influence global education models. Through the coordinated efforts of government, academia, institutions, and industry, major progress has been achieved in implementing ICT in higher education, but some major challenges, elaborated below, must be addressed due to the rapid changes inherent in new technologies.

Lack of scientific theoretical framework to guide the construction of innovative education systems. Due to the lack of integration among the elements of the teaching and learning system, such as resources, physical spaces, technology platforms, data, management, services, and ecosystems, there is an urgent need for scientific theoretical models that support the development of longitudinal and large-scale blended teaching and learning systems that have the same effectiveness as traditional educational settings. This is necessary to realize intelligent decision-making in higher education, promote instruction innovation, and improve the efficiency and effectiveness of teaching and learning.

Lack of integrated intelligent solutions for whole-process and cross-sector management, all-round teacher service, and full-chain infrastructure systems that can ensure the quality of large-scale online or hybrid education systems. A full-spectrum, cross-sector education system has not been established. Such a system would integrate enrollment, entry assessment, orientation, course registration, course design, teaching delivery, examination, thesis defense, and graduation into one system, facilitating the organization, delivery, and evaluation of online teaching and helping learners to carry out efficient and high-quality online learning activities. ICT-based teaching support services such as teacher training, instruction design, and online Q&A have not yet been developed, so it is difficult to quickly assist teachers seeking to improve their information literacy, initiate teaching innovation, or make full use of ICT. A robust, full-chain infrastructure has not yet been established. Such a system could manage computing, storage, networks, and information flow, supporting large-scale online teaching and learning activities.

Lack of coordinated ecosystem that can sustain the continuous evolution of ICT-enabled education. Constructing an innovative education system requires a huge investment in hardware and software products, advanced technology, and professional human resources. A collective and collaborative ecosystem that integrates education, research, and industry is needed to truly meet the growing demands of the rapidly evolving technological era.

Analyses of both research in learning sciences and educational practices grounded in the research reveal the pathway to address these challenges. Tondeur et al. (2021) made the following observations in their review of relevant conceptual models used in research and practice:

Some models have stimulated international research initiatives such as the TPACK (Technological Pedagogical Content Knowledge) Model (Koehler & Mishra, 2009; Mishra & Koehler, 2006); the “Will-Skill-Tool” or “Will-Skill-Tool-Pedagogy” Model (Christensen & Knezek, 2001; Knezek & Christensen, 2016); or pedagogical adaptations of generic Technology Acceptance Models (Davis, 1989; Venkatesh et al., 2003). Other models have been applied to practice such as the Substitution-Augmentation-Modification-Redefinition Model (SAMR; Puentedura, 2006, 2012), the Technology Integration Matrix (TIM; Welsh et al., 2011), the Technology Integration Planning Model (TIP; Roblyer & Doering, 2013), or the Four in Balance Model (Kennisnet, 2014).

These models can be seen as simplified versions of theories that can be used in practical applications or as condensed “wisdoms of practice” that need to be further investigated (Tondeur et al., 2021). Although these models have shed light on how to effectively use technology in education, there have been few attempts to combine them into an overarching theoretical framework. This paper argues that learning engineering could provide a framework for enhancing learning and teaching through the use of technology and addressing the above challenges.

Learning Engineering and Exemplary Practice

Learning Engineering

Given the definitions of learning engineering discussed, it is not difficult to identify the core attributes of learning engineering: theoretical grounding in science, holistic engineering methodology, and inclusive, pragmatic ecosystems.

Theoretical Grounding

Learning engineering applies principles derived from the learning sciences. Its theoretical foundations are a blend of scientific disciplines, including learning sciences, data science, computer science, and instructional system design (Wagner & Lis, 2018). The learning sciences include neuroscience, a branch of biology that studies the nervous system, neurons, and the behavior of the brain; cognitive psychology, which is the study of the human mind through observable behavior; and education research, which focuses on models and interventions at the classroom level and above (Erb et al., 2019). Just as civil engineering, mechanical engineering, and electrical engineering have led to better practices across society, learning engineering can combine the power and discipline of both engineering and the learning sciences to develop better, more reliable, and more effective technological tools for instructors and learners (Wagner & Lis, 2018).

Holistic Methodologies

Engineering mindsets and concepts are reflected in learning engineering. Engineering consists of systematic problem-solving processes. Learning engineering adopts engineering mindsets such as a system perspective that encompasses the full learning cycle: learning theories are used to construct learning models; learning models scaffold instructional design; instructional design directs learning practices; learning practices generate data; and data enrich learning theories.

Engineering also features scalable solutions. Science aims to discover truth, while engineering aims to create scalable solutions to problems that will work in a range of conditions. Currently, the science of learning is not being applied at scale. Therefore, we see the need for a new field that will produce scalable solutions. Scalable complex systems are achieved in part by decomposing systems into modules and creating interfaces between those modules. Interoperability is improved by using standard interfaces (Goodell & Kolodner, 2023). A learning engineering approach can achieve scalable solutions by creating systems that link divergent modules (i.e., learning management units, talent training, services in different dimensions, and interconnected infrastructure).

Inclusive Ecosystem

Learning engineering is a “team sport” requiring multidisciplinary expertise with enough shared understanding for effective collaboration. It involves experts in learning sciences, data science, software engineering, learning experience design, learning environment engineering, learning assessment, measurement, and evaluation, subject-matter expertise, and education and training professionals (Goodell & Kolodner, 2023). Massachusetts Institute of Technology calls for a coordinated research agenda that integrates insights across disciplines that impact learning: “In order to facilitate design of effective solutions, researchers from across the many fields related to education will need to work together—from the social scientists who study impact of education on social systems, to the researchers who explore pedagogical approaches and classroom structures, to the psychologists who study behavior and the neuroscientists who study learning processes in brains” (Willcox et al., 2016). A common research agenda that integrates findings from all fields of education could lead to powerful new insights. It would help build a community of versatile experts who can apply key findings to learning practices across higher education.

Learning engineering is also a “team sport” in that it requires integrated efforts across different types of institutions and among learners, instructors, researchers, government policymakers, and ed-tech providers (Figure 2).

Figure 2.

Attributes of Learning Engineering.

Simon Initiative

Robust Theoretical Framework

The Simon Initiative does not rely on a single theoretical framework or even a singular field. Instead, it builds on a growing body of theoretical principles that support robust learning and instruction. The Initiative provides learning scientists with the results of research from different fields, using, as much as possible, the same core terminology. It is committed to building a community by establishing and supporting collaboration among current and future researchers, computer scientists, and instructors. While a single theory of learning may eventually emerge, the immediate goal is to encourage researchers to maximize the overlap between theories and help the field move beyond the “Toothbrush Problem” (retrieved from LearnLab).

At the Simon Initiative, relevant theories have been developed for understanding different aspects of learning: cognitive factors (understanding cognitive learning—changes in knowledge—that result from instructional events); social and communicative factors (understanding communication as a core enabler of robust learning and how learning occurs through social-communicative interaction); metacognition and motivation (understanding how metacognitive processes and motivation interact with learner factors to influence robust student learning outcomes, and testing the impact of student learning environments); and computational modeling and data mining (using data to advance precise computational theories of how students learn) (retrieved from LearnLab).

Holistic Approach at the Macro Level

The Simon Initiative supports flagship projects spanning every stage of the learning engineering cycle, with LearnLab as the hub of theoretical research, the Open Learning Initiative (OLI) as the hub of learning practices, and the DataLab as the hub of practical research, with the findings channeled back into the LearnLab.

LearnLab. This is the scientific arm of the Simon Initiative. Originally funded by the National Science Foundation, it leverages cognitive theory and computational modeling to identify the instructional conditions that lead to robust student learning. LearnLab's two main goals are to enhance the scientific understanding of robust learning in educational settings and to create a research facility to support field-based experimentation, data collection, and data mining. LearnLab is advancing both basic research on learning in knowledge-rich settings and applied research that contributes to the design and engineering of educational approaches that will have a broad and lasting effect on student achievement. LearnLab enables technologies that support learning analytics and educational data mining, the development of intelligent tutoring systems, online course development, and computer-supported collaborative learning.

OLI. This initiative offers textbook-replacement course content built upon principles gleaned from decades of research in three of CMU's strengths: cognitive science, computer engineering, and human–computer interaction. The OLI provides a harmonious platform for delivering high-quality materials that facilitate groundbreaking research in technology-enhanced learning (TEL), data science, learning behavior, and other areas. It uses open, high-quality courses, continuous feedback, and research to improve learning and transform higher education. Educators and researchers can employ OLI technology in the following tasks:

teach any course in the diverse OLI catalog, and monitor the students’ learning;

build and deliver a new course or an entire program, harnessing the OLI methodology that measurably improves student learning;

convert their existing instruction to the OLI framework, for meaningful data collection and iterative improvements, at scale;

partner with other users to integrate new technologies into instruction, collaboration, or data science research;

provide a testing ground for grant-funded research; or

combine multiple Simon Initiative partner products to tackle challenges in instruction.

DataLab. This lab is the world's largest bank of educational technology data. It includes detailed data about how people learn and how to design and deploy effective learning software. DataLab offers tools for collecting, analyzing, and securely storing data from both offline and online classes. These efforts help educators, researchers, course designers, and partners to improve student learning outcomes by better assessing learner performance, designing courses, testing new ideas, and incorporating insights into teaching practices.

Construction of a Learning Engineering Ecosystem

The goal of the Simon Initiative is to build a learning engineering ecosystem at CMU; its core mission is to make this ecosystem an open global community. Internally, it strives to create the technological infrastructure and human support needed for faculty to feasibly use learning science research to improve their educational practice, instrumentalize their educational practice so that they have access to useful data, and analyze those data to improve student learning outcomes. Globally, it strives to provide accessible tools and methods that any person or institution can adopt and use to improve the outcomes of their own learners.

Learning@ZJU Initiative

Learning@ZJU is a major initiative launched by Zhejiang University (ZJU) based on a down-to-earth learning engineering approach. It addresses the aforementioned challenges faced by ZJU and the wider landscape of higher education in China.

As discussed at the World Artificial Intelligence Conference, intelligence enhancement technology is speeding up, with artificial intelligence (AI) technologies driving the change from Education 1.0 to Learning 2.0. The application of AI will accelerate the development of lifelong learning (Zhang et al., 2020). Conference participants also discussed a new four-dimensional concept of the education landscape consisting of the human world, physical world, technology world, and information world. What is the relationship between this four-dimensional world and Learning 2.0?

First, when teachers and students (who make up the human world) enter a classroom (the physical world), the human and physical worlds interact. Prior to the existence of ICT, teachers imparted knowledge, and students learned knowledge in such environments. When live recording tools are installed in the classroom, these tools (the technology world) change the learning system by introducing recorded lessons and video lessons. For example, in a classroom where voice recognition technology interacts with live broadcasts, it forms a text that can be displayed on a large screen, allowing the teacher's words to be quickly displayed to the students. This accelerates students’ understanding of the knowledge and increases efficiency. At the same time, this text becomes new content input for the knowledge graph; when the video course is combined with the text, it forms a video course with subtitles that can be stored in the course cloud. Mixing and cutting technologies can then be used to transform video courses with subtitles into a series of shorter courses. For example, a 45-min course can be transformed into a number of 5-min courses that are each stored on the ZJU Learning Online platform. These platforms and these courses become part of the information world.

The four worlds interact continuously, producing more content over time. For example, student interaction with the ZJU Learning Online platform can produce student evaluations and personal portraits of students, which can provide digital files that can be used by students seeking employment. After graduation, students can continue to use the ZJU Learning Online platform, which supports lifelong learning. The new content generated by interactions between classrooms, the cloud service, and the ZJU Learning Online platform can be deconstructed or reorganized using the knowledge graph. This forms a Learning 2.0 space with “three links and one core,” that is, the basis for the K-CPS framework developed at ZJU. This framework is described in the following section.

Theoretical Grounding

The Learning@ZJU Initiative takes the Knowledge graph as its core; this core has three arms, smart Classrooms, learning Platforms (e.g., ZJU Learning Online), and cloud Services (e.g., Zhiyun Class), which provide pathways around the roadblocks created by the challenges discussed above. The result is robust learning and effective instruction. This K-CPS framework serves as the theoretical basis of the Learning@ZJU Initiative, built on the learning engineering approach. It provides a general model or guidelines, crafted by ZJU, for the design and planning of ICT-enabled learning environments (Figure 3).

Figure 3.

The K-CPS Framework.

Challenge Addressed: Lack of Scientific Theoretical Framework to Guide the Construction of an Innovative Education System

The K-CPS framework bridges the gaps between classrooms, courses, and teaching platforms. Each element functions independently and in connection with the other elements. By itself, the ZJU Learning Online platform supports course preparation, course delivery, live broadcasts, interactions, and examination and evaluation; externally, it pushes educational data and content into the knowledge graph and the Zhiyun Class cloud service, respectively. The smart classroom component independently collects educational data such as classroom environment, instruction videos, interactions, and instruction content from each smart classroom and then transmits the data to the knowledge graph and Zhiyun Class. The Zhiyun Class cloud service independently conducts intelligent processing tasks such as intelligent audit, voice recognition, simultaneous interpretation, and deep labeling and classifies the educational resources and data pushed by the ZJU Learning Online platform and smart classroom component; externally, it pushes the processed data to the knowledge graph. The knowledge graph independently reconstructs and analyses the data collected by the other components, and pushes the tailored learning schemes to the ZJU Learning Online platform, where they become accessible to students. Together, these components create an ICT-enhanced learning environment.

Smart classrooms not only support classroom teaching but also collect educational data and content through methods such as IoT perception, identity recognition, and live and recorded instruction videos. In the hybrid environment designed using the K-CPS framework, educational data and videos recording classroom performance, teacher–student interactions, and learning trajectories are automatically collected using devices such as classroom interaction apps, smart cameras, and intelligent recording and broadcasting equipment. The educational data generated in these smart classrooms is transmitted to the knowledge graph, which facilitates learning analysis and supports the generation of personalized learning plans. The teaching videos are pushed to Zhiyun Class to enrich the teaching resource library.

The ZJU Learning Online platform uses AI, big data, and learning measurement technology to realize the following goals: cross-campus distribution of teaching through its video-on-demand and classroom interaction functions; teaching available anytime and anywhere through a mobile app; blended teaching enabling online self-regulated learning and offline classroom interaction; and visual statistical reports that summarize students’ learning status. In the visual analysis report, teachers can view users’ access of various resources such as courseware and videos, the completion status of student assignments and tests, and grade statistics. Furthermore, teachers can also follow students’ progress in the course, which allows precise adjustments to teaching strategies.

The cloud service (Zhiyun Class) not only stores all of the recorded offline course videos for students to review but also offers live teaching broadcasts with classroom interactions for online students. Additionally, serving as a vast repository of teaching video resources, Zhiyun Class uses intelligent processes such as intelligent review, voice recognition, simultaneous interpretation, and advanced tagging to categorize resources and data received from the ZJU Learning Online platform and smart classrooms. It then pushes these data to the knowledge graph, enabling the system to swiftly construct subject-specific video knowledge graphs, simplifying large-scale personalized online learning for students. Every day, Zhiyun Class aggregates all of the outputs of all of the courses taught in the whole school and makes them available to students through online live streaming and recorded replays. Additionally, features such as PowerPoint presentations, interactive student discussions, simultaneous interpretation in 28 languages, AI-powered note taking, and keyword analysis are integrated into both live streams and recorded replays, achieving a true blended teaching and learning.

The knowledge graph, as the core of the K-CPS framework, functions as an intelligent brain that automatically gathers extensive teaching and knowledge data from the smart classrooms, ZJU Learning Online platform, and Zhiyun Class cloud. Using the automatically generated knowledge graphs, the system intelligently analyzes teaching data and makes personalized matches between teaching resources and teachers and students, creating individualized teaching and learning schemes that are disseminated on the ZJU Learning Online platform. This facilitates large-scale personalized online and offline education for teachers and students, ultimately achieving a student-centered approach to instruction.

Holistic Approach at the Micro Level

Challenge addressed: Lack of integrated intelligent solutions for whole-process and cross-sector management, all-round teacher service, and full-chain infrastructure systems that can ensure the quality of large-scale online or hybrid education systems.

A holistic learning engineering approach is needed to link the components distributed across the full spectrum of education. The Learning@ZJU Initiative uses this approach to achieve one-stop management, whole-process support, all-round training, and full-chain infrastructure support.

One-stop management and whole-process support. The Learning@ZJU Initiative features smooth data circulation and intelligent data analysis. It uses unified identity authentication to link distributed platforms, including the educational administration system, student affairs system, office system, and personnel management system, creating a one-stop integrated management service for preclass organization, in-class instruction, and off-class evaluation—a complete closed loop of online and offline management that reduces the need for teachers and students to shuttle between multiple management platforms and reduces the complexity of innovative teaching initiatives. Learning@ZJU Initiative enables holistic online education that covers the whole education process, from enrollment to orientation, admission tests, instruction and learning, exams, thesis defense, graduation, and finally employment. It also lays a solid foundation for future-oriented open-loop education systems.

All-round training. The cornerstone of the system is the training offered by the ZJU DingTalk service, which makes a series of applied training opportunities available in the form of live broadcasts, video conferences, and face-to-face instruction. This increases the ease of use for both teachers and students.

Full-chain infrastructure. Together, the school's network infrastructure, service computing resources, network security, large screen displays, and other resources provide full-chain infrastructure support from the users’ end to the servers’ end, providing everything from regular recording to intelligent processing and streamlined distribution of course videos.

An Open-Loop Ecosystem

Challenge Addressed: Lack of Coordinated Ecosystem That can Sustain the Continuous Evolution of ICT-Enabled Education

The Learning@ZJU Initiative is committed to integrating efforts across units within the university and across sectors beyond education. As mentioned in the earlier section, building an innovative education system requires a huge investment in hardware and software, advanced technologies, and professional human resources.

Internally within the university. The Learning@ZJU Initiative is cooperation between different business units, which execute their own functions in a cooperative manner. This creates a five-dimensional system that, through the integration of “regulation, honor, service, technology and exchange,” motivates teachers to engage in instructional reform, improve the information literacy of stakeholders, and stimulate educational innovation.

Externally beyond the education sector. The Learning@ZJU Initiative has built an open-loop ecosystem that is continuously changing. Building on the concepts of crowdfunding, collective intelligence, and coconstruction and treating digital intelligence technology as the driving force of systematic reforms in education, the Learning@ZJU Initiative cooperates with top industry partners, sister universities, and government policymakers with the aim of forming an open-loop ecosystem that connects government, industry, university, and research partners and creates a base for research on learning engineering and the development of educational technology that will empower the sustainable development of ICT-enabled education.

Summary

Learning Engineering Approach and the Learning@ZJU Initiative

This section summarizes how the learning engineering approach can solve the challenges facing ICT-enabled education, using the Learning@ZJU Initiative and Simon Initiative as examples (Figure 4).

Figure 4.

Applying the Learning Engineering Approach to the Learning@ZJU Initiative.

Simon Initiative vs. Learning@ZJU Initiative

Both the Simon Initiative and the Learning@ZJU Initiative exemplify the application of learning engineering to ICT-enabled education, although they have different theoretical foundations, adopt different holistic approaches, and build distinct ecosystems. A comparison between the two has implications for the implementation of learning engineering in ICT-enabled education.

The Simon Initiative is grounded in a robust theoretical framework consisting of cognitive factors, social and communicative factors, metacognition and motivation, and computational modeling and data mining. Its holistic approach uses a macro perspective that encompasses the full learning cycle and it aims to construct a learning engineering ecosystem within the university and an open global community.

The Learning@ZJU Initiative is grounded in the K-CPS framework, adopts a holistic approach with a micro perspective that integrates divergent modules and constructs an open-loop ecosystem within the university and beyond the education sector (Figure 5).

Figure 5.

Two Examples of a Learning Engineering Approach Applied to Education.

Current Outcomes of the Learning@ZJU Initiative

Significant Enhancement of Students’ Digital Learning Literacy

The proportion of students actively using the ZJU Learning Online platform is 100%, and the frequency of their use of the platform is continuously increasing. The number of daily average visits exceeds 220,000. Given the 330,000 course interactions and 278,000 views of classroom recordings, digital learning can be seen as the norm. Over the same period, 1,740 randomly selected courses were compared: 70% showed improvement in average grades, 76% experienced a significant increase in pass rates, and 73% saw a marked rise in excellence rates. Furthermore, the platform's online employment, online guidance, and online service components have shared information on over 225,000 positions, resulting in a year-on-year increase of 77.24% in the employment of ZJU graduates at national research institutes.

Significant Enhancement of Teachers’ Digital Teaching Skills

The ZJU Learning Online platform encompasses more than 50,000 online courses. It has facilitated the establishment of 1,000 top-tier MOOCs, 330 high-quality MOOCs, and 240 national-level high-quality open courses. In addition, 160 seminars and training sessions on educational reform have been organized for teachers. A survey indicated that 100% of the teachers use the ZJU Learning Online platform and Zhiyun Class cloud service in their teaching. Through diverse and innovative teaching practices, teachers have substantially enhanced their information literacy and teaching skills.

Resource Aggregation and Application

The ZJU Learning Online platform has garnered over 250 million visits, generated an excess of 1.45 million intelligent keywords, and extracted over 22.5 million pages of PowerPoint content. Additionally, it has facilitated more than 780,000 h of speech recognition. Over 800 smart classrooms have been constructed on ZJU campuses, which host an average of more than 1,200 h of live courses per day and have supported teachers and students in conducting over one million live broadcasts/conferences (equivalent to more than 15 million hours). The user base spans 149 countries and regions, enabling resource sharing beyond geographical boundaries.

Social Impact

The Chinese Ministry of Education nominated the Learning@ZJU Initiative for the UNESCO King Hamad Prize. It has also been recognized as a digital reform achievement in Zhejiang Province and supported the development of the Zhejiang Graduate Union MOOC platform “Learning in Zhejiang” and a global summer school program spanning 81 countries and regions. The model has been replicated in over 30 universities, expanding its impact to the global scale.

Discussion

Learning engineering is drawing increasing attention from academics, industry professionals, and governments. This paper presents a holistic case study of the application of the learning engineering approach to ICT-enabled education at a university in China. However, the effectiveness of such initiatives takes time to validate, and further research is necessary to assess whether TEL has been achieved and to investigate the adoption of TEL resources.

Evidence on the effectiveness of educational technology is required. Robust learning is characterized by long-term retention, is transferable to new situations that differ from the learning situation in various ways, and accelerates future learning in new situations (Koedinger et al., 2012). Computer-based learning environments offer unprecedented opportunities to gather fine-grained, large-scale, longitudinal data on learning in an unobtrusive manner (Azevedo & Aleven, 2013). In the case of the Learning@ZJU Initiative, Zhou et al. (2023) explored a model for evaluating online learning processes that is based on the quantitative analysis of learning behavior data. Chen (2022) conducted research on teaching argumentation in blended learning environment using data from the ZJU Learning Online platform and concluded that the platform broadens students’ engagement. Further evidence-based studies of the effectiveness of Learning@ZJU Initiative are expected in the future, and the feedback they provide will further enhance the platform.

Research on how to encourage the use of effective learning technologies and to stimulate innovation is required. Kirkwood and Price (2014) adopted HEFCE's three-part definition of TEL: Efficiency (conducting existing processes in a more cost-effective, time-effective, sustainable, or scalable manner); Enhancement (improving existing processes and outcomes); and Transformation (radical, positive change in existing processes, or the introduction of new processes). However, Tondeur et al. (2021) implied that although for decades scholars such as Hattie (2009) have affirmed the findings of Kulik and Kulik (1991) that technology can effectively promote learning, Schneider and Prekel's (2017) summary of 38 meta-analyses suggested that technology affordances still lag behind five other categories of instructional practices in promoting higher education achievement. This echoes the findings of Tondeur et al. (2008), who concluded that most technology affordances do not exploit the full potential of IT; in fact, they use only its basic features. Carnegie Mellon University, a university with a particularly rich history in TEL and other instructional innovations, has also experienced a surprisingly slow rate of adoption of many of these demonstrably effective innovations (Smith & Herckis, 2018). Smith and Herckis (2018) used a range of research methods, including faculty surveys, ethnographic studies, and semistructured interviews with both faculty and relevant professional staff, to identify the barriers to the adoption of these technologies and recommendations to encourage dynamic efforts to implement TEL-based instructional innovations. We expect that similar studies will be conducted using data from the Learning@ZJU Initiative.

Conclusion

This paper presents a comprehensive case study of the application of the learning engineering approach in the Learning@ZJU Initiative, demonstrating its potential to drive systemic innovation in ICT-enabled education. This initiative integrates theoretical foundations, holistic system design, and cross-sector collaboration, creating a promising model for scalable and inclusive learning ecosystems. While early outcomes have indicated increased student engagement and richer data-driven insights into learning behaviors, the long-term effectiveness of such initiatives requires rigorous evidence-based evaluation. Future research should focus on validating the sustained impact of TEL practices on learning retention, transfer, and accelerated learning. Moreover, developing strategies to encourage the broader adoption of evidence-based, effective educational technologies remains a critical task. As past studies have shown, technological affordances often fall short of their potential due to a lack of targeted implementation support and institutional commitment. The Learning@ZJU Initiative offers fertile ground for addressing these challenges through its use of continuous feedback and iterative improvement. Ultimately, advancing learning engineering as a discipline will depend on sustained cross-disciplinary collaboration, robust empirical research, and responsive educational ecosystems.