Everything about our life is complex. It should not be so, but this is the reality now. Perhaps, the blame for this exponential rise in the complexity of life should be on us. This is because of our increasing desire to do more, produce more, and innovate more in each and every moment of our life. We wish to find new ways of living, we wish to find new planets for living, and all this has increased the complexity of everything we do. Such increasing complexity requires new ways of “governing” our life and the various complex systems around us that support us to live a complex, “modern” life.

The word “governance” is increasingly common in many sectors and contexts these days, corporate governance, global governance, transnational governance, urban governance, risk governance, sustainable governance, and more. Bevir (2011) states, “at the most general level, governance refers to theories and issues of social coordination and the nature of all patterns of rule”. The traditional methods and modes of governance are considered to be monolithic, formal, and hierarchical and are no more seen as sustainable for managing the complex world of today. The contemporary use of the word “governance” refers to and emphasizes “various new theories and practices of governing and the dilemmas to which they give rise” and “draws attention to the complex processes and interactions that constitute patterns of rule” (Bevir, 2011). Governance involves in its decision-making process institutions that are not necessarily from the government (in its traditional, formal, sense) but include various stakeholders from the public, industry, voluntary sectors, and others. Therefore, decisions are made both autonomously and collectively by different actors, rendering the boundaries and responsibilities of governance to become blurred. Participatory governance has become common because “the formal state apparatus has frequently proved to be inadequate in dealing with growing social complexities” and “actors from the voluntary and private sectors have become increasingly involved in the governance process” (Chhotray and Stoker, 2009). The government is more seen as a facilitator, integrator, or coordinator, that is “able to use new tools and techniques to steer and guide” (Stoker, 1998). The asynchronous nature and blurriness of responsibilities of contemporary governance, hence, have given rise to several dilemmas including the difficulties associated with blame avoidance and accountability and risks of governance failure, among others (Stoker, 1998).

A critically important area where the role of governance is crucial is the education sector (Burns and Köster, 2016). Education governance has been a subject of much investigation, debate, controversy, and polyvalence (Wilkins and Olmedo, 2019). This is due to the complex and the rapidly evolving nature of the education sector involving many government levels (national, regional, local, etc.) and stakeholders including all public segments, and more importantly, due to the role that education plays in developing theories, practices, and social and economic capacities for all forms, disciplines, and domains of governance. The education sector is going through rapid transformations due to social, economic, and technological reasons such as the availability of free and paid massive open online courses (MOOCs), the industry gradually prioritizing labor skills over university degrees, the education sector facing new economic challenges, the increasing demands by modern learners for any time, anywhere learning, the need of educational institutions to keep abreast of the rapidly changing technology and knowledge and accordingly adjust their content and content delivery systems up to date, the need to use many new sources of data in education-related decision making, and more (Mehmood et al., 2017a). The changes in global political ideologies and environments have also transformed education governance, making it a decentralized, complex, multi-level, and multi-stakeholder sphere subject to ad hoc, autonomous, and participatory decision-making. Education governance is seen as polyvalent because it is “experienced by some as a part of the dangerous and mischievous practices of the ‘hidden hand' of the market or neoliberalism more generally, while those who fear the tyranny of hierarchies and are distrustful of top-down systems celebrate it as an empowering tool for democratic change, innovation and improved effectiveness or transparency” (Wilkins and Olmedo, 2019). Saguin (2019) emphasizes that education governance “should be seen as establishing a specific type of relationship between governmental and non-governmental actors that often involves a mix of legal, market, network, and corporatized approaches”. Researchers have also warned against the risks of governance failures in education that can “occur when private actors ‘capture' the government to channel public resources to pursuing private interests” (Saguin, 2019). The complexity of education governance has also led to the development of new disciplines such as the sociology of disciplines that studies the effects of public institutions and individual experiences on education and its outcomes (Raab, 1994; Davies and Guppy, 2014).

Knowledge and human and social capital are fundamental to human progress. However, the mainstream role of education is increasingly seen and adopted as “education for employment”. Rather, the role of education should extend to “the notion of social and collaborative governance where the society will collaborate to train each other in maintaining its knowledge, moral fiber, good practice, resilience, competitiveness, and for bringing innovation and becoming a knowledge-based economy” (Mehmood et al., 2017a). This will require a converged ubiquitous infrastructure comprising many systems that allow information sharing, collaboration, and decision-making between different stakeholders, as necessary, for defined objectives including the triple bottom line (TBL).

The challenges facing governance have been further exacerbated by the COVID-19 pandemic that has affected many things in our lives including restrictions on being in physical spaces leading to our accelerated transition to being in digital spaces. Governance for many sectors has been affected, including the education sector with teaching and learning moving to and from between online and physical learning spaces. These pandemic-related challenges are here to stay and, therefore, new approaches are needed to address the governance-related challenges. Both theoretical and practical contributions are required from the community to education and general governance.

The complex challenges mentioned above, we believe, can be addressed by smartization of our societies and living spaces, enabling a sustainable future for us humans, due to its data-driven analytics approach and its focus on TBL—social, environmental, and economic sustainability. Precisely, smartization relies on collecting data and making informed decisions on policy and action using cutting-edge technologies such as the internet of things (IoT), big data, artificial intelligence, and large-scale distributed computing. Our research aimed to explore this smartization concept and develop methods based on data-driven analytics for informed decision-making and governance.

To accomplish this objective, this study provides a case study on sustainable participatory governance using a data-driven discovery of parameters for planning online, in-class, and blended learning in Saudi Arabia evidenced during the COVID-19 pandemic. The data-driven parameter discovery approach we propose is generic and can be applied to the governance of any sector. The specific case study on education is used to elaborate on the proposed approach. A high-level framework depicting our proposed data-driven approach for participatory education governance is provided in Figure 1. It shows that various actors work to achieve certain educational objectives using a range of data sources. The triple bottom line is shown at the bottom that could be considered as the high-level or ultimate objectives of the education governance. Examples of actors include people, government, the voluntary sector, social enterprises, and others. The data sources include, without the intention to be exhaustive, -social media, government, institutional, academic and scientific literature, industrial, and other data. The data sources are modeled and analyzed using machine learning to discover education governance parameters and macro parameters. The parameters are evaluated, validated, and improved using national and international good governance parameters. Good governance parameters, for instance, can also be included in the design of the data-driven governance parameter discovery system, such that automated comparison and evaluation of a government's services are made against established standards of good governance. The good governance parameters themselves can be improved using the data-driven self-improving approach. The improved education governance parameters are applied and tested in real environments that create effects on data sources, objectives, actors, and sustainability. This process is repeated iteratively and continuously to dynamically improve education governance. The government could use the knowledge gained through this data-driven discovery process to develop better governance apparatuses, and this whole process could be implemented as a perpetual loop for real-time or dynamic governance with much finer levels of engagement with the public and other stakeholders.

To investigate the proposed approach and framework, we developed a software tool comprising a complete machine learning pipeline and used a dataset comprising around 2 million tweets (1,878,993 tweets, to be precise) in the Arabic language collected during the 14-month period, i.e., 1 October 2020 to 17 December 2021. We discover fourteen governance parameters including, among others, Pandemic Measures, Digital Services, Nurturing Positive Behavior, Exam Procedures, and International Collaborations. These fourteen parameters are grouped into four governance macro-parameters, namely, Success Factors, Challenges, Socioeconomic Sustainability, and Accountability and Transparency.

This study makes important and significant theoretical and practical contributions to the literature. First, we proposed a novel data-driven parameter discovery approach for governance applicable to any sector, and, to elaborate on the proposed approach, we provided a case study in education governance. A theoretical framework for data-driven participatory education governance is developed and provided in this work. Second, we developed a complete case study and software for the data-driven discovery of parameters from social media (Twitter) data for planning and governance of learning during COVID-19. The practical contributions that we provide in this paper are novel, and completely different from any of the other works because it attempts to comprehensively grasp and discover governance parameters related to the education sector from Twitter data using unsupervised machine learning. Third, the research provides evidence to support the general literature on data-driven smart cities research and reinforces that policy and action on smart cities should be supported with data (Liu et al., 2017; Yigitcanlar et al., 2020; Bibri, 2021a) and that social and digital media provides a convenient and important source of such data (Alomari et al., 2021b). The literature review section discusses the related works and establishes the research gap in detail.

This work is an extension of our earlier work that had considered 128 thousand tweets for a 66-days period from October 2020 to December 2020 (Alswedani et al., 2022). The work presented in this study makes substantial improvements to the data collection process and overall methodology, including the collection of around 2 million tweets for a period of 14 months (October 2020- December 2021). The period has undergone significant public and governance-related changes due to the evolving situation of the pandemic and education, and therefore, this study reports new parameters, insights, and findings compared to the previous paper. Also, it extends the data analysis content presented in the earlier work. We have performed a temporal analysis of each parameter and provided visualizations for all the parameters. Moreover, we provide in this study a theoretical framework for data-driven governance.

The particular aspect in which we propose the use of machine learning for urban planning and governance is expected to be a trendsetter for future works and provide a significant impact on research and practice in this area. The discovered parameters by the tool demonstrate the possibility and benefits of our sustainable participatory planning and governance approach, allowing the discovery and grasp of important dimensions of the education sector in Saudi Arabia, the complexity of the policy, procedural and practical issues in continuing learning during the pandemic, the factors that have contributed to the success of teaching and learning during the pandemic times, both its transition to online learning and its return to in-class learning, the challenges public and government have faced related to learning during the pandemic times, and the new opportunities for social, economic and environmental benefits that can be drawn out of the situation created by the pandemic. The parameters and information learned through the tool can allow governments to have a participatory approach to governance and improve their policies, procedures, and practices, perpetually through public and stakeholder feedback.

Note that, in this study, we made contextual translations of the Arabic words and tweets for the sake of the English reader. The tweets in many cases were not provided in full to preserve the privacy of the tweeters.

The rest of the article is organized as follows. Section Literature Review provides a review of the relevant literature. Section Methodology and Design introduces the methodology and describes the tool design. Section Parameter Analysis and Findings explains the discovered parameters. Section Discussion gives additional insight into the results. Section Conclusions and Future Work concludes the paper.

We provide here in this section a review of works related to the study. The focus of the study is Twitter data analytics for urban governance of the education sector during the COVID-19 pandemic; hence, we first review works on smart cities followed by reviews of works on education and smart cities, social media analytics in general and during COVID-19, Twitter data analytics in education, and topic modeling in Twitter data analytics.

Smart societies and cities are driven by objectives such as triple bottom line, innovation, and efficiency optimization through data-driven computing (Bulkeley and Castán Broto, 2013; Mehmood et al., 2017a; Ahmad and Mehmood, 2020; Cugurullo, 2020; Yigitcanlar et al., 2020; Bibri, 2021b). Data-driven machine learning is at the heart of it (Yigitcanlar et al., 2021b). The terms such as “data-driven”, “data-intensive”, etc., are used with machine learning to highlight that machine learning methods are driven by or use large amounts of data, see e.g., (Jordan and Mitchell, 2015) for machine learning trends, perspectives, and prospects. Many bodies of literature have been proposed to investigate and advance data-driven computing for cities and urbanization such as the use of big data for planning smart cities (Batty, 2013), sustainable, smart, resilient, low carbon, knowledge cities (De Jong et al., 2015), urban computing (Zheng, 2017), the cases of London and Barcelona as data-driven smart cities (Bibri and Krogstie, 2020), energy planning with big data (Estiri, 2017), intelligent urban computing with big data (Liu et al., 2017), geospatial data in urban studies (Sun and Du, 2017), and datafication of urban society (Calvo, 2020). Several studies have sought to contribute to the computational aspects of big data analytics to improve the speed of analytics such as road traffic prediction (Aqib et al., 2019b), metro urban transportation systems (Aqib et al., 2019a), detection of diseases in cities (Alotaibi et al., 2020), logistics (Suma et al., 2017), detection of events for urban management (Alomari et al., 2019), collaboration among autonomous vehicles (Alam et al., 2019), and more.

The education sector is going through rapid transformations due to social, economic, and technological reasons. There is an increasing interest in education in smart cities. Molnar (2021) discussed the shortcomings of educational initiatives in smart cities and how to address them. Sadeh et al. (2020) looked into the technological aspects of learning in smart cities. Kwet and Prinsloo (2020) examined developments in the “smart classroom” for higher education. They present the scope and limitations of smart classrooms in the context of university initiatives in smart cities. Some researchers have looked into proposing educational applications with immersive technologies. For example, Gnanadurai et al. (2022) proposed a learning system (AstroSolar) with immersive technology to help students in the primary school develop a better understanding of the solar system. They utilized augmented reality (AR) and virtual reality (VR) to combine the physical environment with digital information. Many VR-based tools were proposed to help students with various subjects in space (Mintz et al., 2001; Yair et al., 2003), mathematics (Hwang and Hu, 2013), medicine (Elliman et al., 2016), and engineering. Mehmood et al. (2017a) propose teaching and learning big data framework to improve lifelong learning in smart societies. They used eleven widely used datasets to evaluate various functions of the proposed framework. Reviews on education in smart cities can be found, for example, in the studies of Mehmood et al., 2017a and Molnar, 2019.

Many COVID-19-related studies which are utilizing Twitter data have been proposed. Some researchers have explored the thematic analysis of Twitter data. For example, Samuel et al. (2020) analyzed Twitter data to discern sentiment related to the COVID-19 pandemic. The authors extended their previous study of Samuel et al. (2020) to identify the major themes on Twitter posts that are associated with the debate on reopening the US economy during the pandemic. The impact of COVID-19 on society was studied by Duong et al. (2020a,b) using sentiment analysis of Twitter data. Abdulaziz et al. (2021) examined the sentiment of the COVID-19 tweets, as well as the extracted topics. Sentiment analysis was conducted by Abd-Alrazaq et al. (2020) to capture the major trends in Twitter posts related to Coronavirus disease. A number of other studies examined a thematic analysis of COVID-19-related tweets, including Jimenez-Sotomayor et al. (2020) as well as Das and Dutta (2020). A review of some other COVID-19-related studies on social media data analysis including Twitter was published in Alomari et al. (2021a).

Few studies have focused on COVID-19 data analytics using social media in the Arabic language. Essam and Abdo (2021) analyzed Arab communities' reactions to the pandemic of COVID on Twitter. Alsudias and Rayson (2020) analyzed Arabic tweets associated with COVID-19 to detect rumors. Alanazi et al. (2020) studied Arabic tweets to identify the most common syndromes associated with COVID-19 cases and the order in which they appear. Researchers have also analyzed Twitter data in Arabic to examine government measures and public concerns about COVID-19. A sentiment analysis study was presented by Alhajji et al. (2020) to explore the effects of closing public and private facilities in Saudi Arabia in response to the COVID-19 outbreak. In a study published in Alomari et al. (2021a), researchers analyzed Arabic Twitter data to detect public concerns and the Saudi Arabian government's pandemic response to COVID-19. (Alomari et al., 2021a) present a review of other literature related to COVID-19 using Twitter data in Arabic.

Several studies have investigated teaching and learning through the use of social media and machine learning (Alsulami and Mehmood, 2018). For example, an analysis of students' learning using social media data was provided by Verma et al. (2016). An analysis of Twitter data related to engineering students was conducted by Lande and Dalal (2016) to explore issues that affect student learning as well as to help improve the educational system. A number of studies have focused on using social media applications such as Twitter and Facebook to enhance learning (Kechaou et al., 2011; Chen et al., 2014; Verma et al., 2016).

Some researchers have also looked at learning during the COVID-19 pandemic. Some have focused on studying students and their attitudes while others have focused on teachers and other stakeholders. These include the view toward online learning of people in Saudi Arabia (Aljabri et al., 2021), higher education students in Pakistan (Adnan and Anwar, 2020), high school students in Indonesia (Bestiantono et al., 2020), primary school teachers (Rasmitadila et al., 2020), and K-12 teachers (Harron and Liu, 2020).

Several studies have focused on studying Twitter data analytics using topic modeling based on the Latent Dirichlet Allocation (LDA) algorithm. These include the work by Zahra (2020) on finding semantic relations related to COVID-19 in Twitter data, a supervised model for regression and classification by Korshunova et al. (2019), detecting public concerns and pandemic measures from Twitter data (Alomari et al., 2021a) and evaluating LDA performance for sentiment classification by Onan et al. (2016). Alomari et al. (2021a) provided a relatively detailed review of topic modeling using LDA.

The literature review establishes that currently there are only three studies that can be considered remotely related to the practical side of our research due to their focus on sentiment analysis of social media for learning during the COVID-19 pandemic. However, these studies focus on sentiment analysis of specific education issues during the pandemic as opposed to the discovery of education governance parameters provided in our research. These include a study on the pandemic response of university students in the US (Duong et al., 2020a), an exploration of public sentiments on closures of private and public facilities in Saudi Arabia (Alhajji et al., 2020), and an analysis of peoples' sentiments on distance learning in Saudi Arabia using Twitter data (Aljabri et al., 2021). Our research is completely different from any of these studies because of its specific focus and in terms of the dataset (the nature of the data due to data collection parameters, the time period of the collected data, the data size, language, and geography), methodology and design (the overall design of the machine learning method and pipeline), our innovative approach of using AI to discover governance parameters, and our findings (Success Factors, Challenges, Socioeconomic Sustainability, and Accountability and Transparency).

This section describes the design and methodology of our proposed tool. Figure 2 shows the architecture of the proposed system which involves five modules, including data collection and storage, data preprocessing, parameter discovery (Education Governance), validation, and visualization and reporting. Section System Overview presents an overview of the proposed system. The architecture's modules will be presented in Sections Data Collection and Storage, Data Preprocessing, Parameters Discovery (Education Governance), Validation, and Visualization and Reporting. Additional details about the general methodology related to this study can be found in the study of Alswedani et al. (2022).

In this section, we discuss the results of our proposed system. Section Parameters and Macro-Parameters: An Overview provides an overview of the macro-parameters and parameters extracted from Twitter data using LDA. Sections Success Factors, Challenges, Socioeconomic Sustainability, and Accountability and Transparency provide a detailed explanation for every urban governance parameter with examples from the collected data.

We provided a case study on sustainable participatory governance using a data-driven parameter discovery approach for planning online, in-class, and blended learning in Saudi Arabia evidenced during the COVID-19 pandemic. The data-driven parameter discovery approach we propose is generic and can be applied to the governance of any sector. The specific case study is used to elaborate on the proposed approach. We discovered fourteen governance parameters grouped into four governance macro-parameters. The first macro-parameter Success Factors includes the governance parameters that have enabled the continuation and effectiveness of teaching and learning during pandemic times. The success factors included effective implementation of pandemic measures to control the potential increase in infection rates due to in-class learning, activities to increase awareness about the importance of education and to motivate everyone to continue and excel in education during the pandemic, supporting online and hybrid learning through training, praising, and championing teachers and others, high-speed internet access, effective digital learning services and platforms, and positive behavior nurturing such as honesty and community care. To elaborate, for example, the government developed strategies for improving educational outcomes by providing support to students, teachers, families, and other stakeholders. Training courses were provided to equip teachers, students, and other stakeholders with the essential skills to access and use the digital teaching and learning platforms. The Ministry of Education had launched several learning resources including providing free access to lectures through Satellite TV (e.g., Ain Educational Channel) and YouTube. The government's decision to return to in-class learning in physical learning spaces was accompanied by various precautionary measures to ensure the health and safety of students, teachers, and others. For instance, anyone who did not take the required COVID-19 vaccine doses was prohibited from entering the physical learning spaces.

The macro-parameter Challenges embodies issues that have been challenging for people to adopt as well as the government's strategies to help manage the difficulties people have faced. These include the changes in the Saudi education systems developed to improve teaching and learning in the country, the government's decision to shift in-class learning to online learning (in the early days of the pandemic, early 2020) and vice versa (since August 2021), the changes in the specific procedures of exams during the pandemic, and the difficulties and benefits of morning and afternoon school schedules. The third macro-parameter Socioeconomic Sustainability has detected and captured social and economic benefits of the transitions in teaching and learning during (and necessitated by) the pandemic. It was experienced that online learning allows operational efficiencies, allows better accessibility to learn across urban and rural geographies and different social classes, and provided new economic opportunities for people. The fourth macro-parameter Accountability and Transparency includes government activities related to the transparency, international collaboration, and evaluation of educational policies, systems, and practices in Saudi Arabia. These aspects of learning were experienced by tweets on activities such as international visits by foreign education officials, engagement of Saudi students in PIRLS tests through digital platforms, and evaluation of online learning in Saudi Arabia by international organizations. The Parameter analysis and findings section provides a detailed discussion of these issues along with supporting references and tweets.

This study makes important and significant theoretical and practical contributions to the literature. First, we proposed a novel data-driven parameter discovery approach for governance applicable to any sector, with a case study in education governance to elaborate on the proposed approach. A theoretical framework for data-driven participatory education governance was developed and provided in this literature. The contribution of our study to theory is that it provides a framework in which data-driven governance can be developed for any sector. It identifies actors, data sources, and objectives and provides a pathway (comprising machine-learning-based analytics, governance parameter discovery, and improvements in relation to established good governance parameters) to develop better governance parameters and apparatuses. This whole process could be a perpetual loop for real-time or dynamic governance with much finer levels of engagement with the public and other stakeholders.

Second, our research provides novel practical contributions. We developed a complete case study and software for the data-driven discovery of parameters from social media (Twitter) data for planning and governance of learning during COVID-19. The practical contributions that we provide in the present study are novel and completely different from any of the other literature because it attempts to comprehensively grasp and discover governance parameters related to the education sector from Twitter data using unsupervised machine learning. Moreover, the findings of our study (specific parameters, experiences, opinions of people, etc.) provide specific insights into education governance, particularly in a time of disruption (the COVID-19 pandemic).

Third, the research provides evidence to support the general literature on data-driven smart cities research and reinforces that policy and action on smart cities should be supported with data (Liu et al., 2017; Yigitcanlar et al., 2020; Bibri, 2021a) and that social and digital media provides a convenient and important source of such data (Alomari et al., 2021b). The parameters detected by the system clearly show the possibility and benefits of our tool, allowing the discovery and grasp of important dimensions of the education sector in Saudi Arabia, the complexity of the policy, procedural and practical issues in continuing learning during the pandemic, the factors that have contributed to the success of learning during the pandemic times, both its transition to online learning and back to in-class learning, the challenges public and government have faced related to learning during the pandemic times, and the new opportunities for social, economic, and environmental benefits that can be drawn out of the situation created by the pandemic. The parameters and information learned through the tool can allow governments to have a participatory approach toward governance and improve their policies, procedures, and practices perpetually through stakeholder feedback. Earlier studies on data-driven smart cities have discussed the benefits of data-driven approaches in smart city design and operations; however, none of the earlier studies have proposed a theoretical framework to develop data-driven governance similar to ours and practical contributions to how data-driven governance can be built into existing systems. Moreover, none of the existing bodies of literature have provided a theoretical framework and practical system prototype for education governance. Our research, therefore, advances the existing theory and practice on data-driven smart cities.

Certainly, many more efforts are needed by the community to advance what we have proposed here in both the theory and the practice. First, the proposed approach, both its theory and practice, needs to be investigated further. For example, there is a need to understand the type of data sources that can be used for the governance parameter discovery process. What is the relationship between the type of data sources and the discovered parameters? What different analytics methods can be used to discover parameters. What is the relationship and the effect of these discovered parameters with actual governance? What policy and practice apparatuses can be developed from this parameter discovery and improvement processes. While new theories need to be developed for data-driven policy and action development, there is also a need to look at how these new developments and theories contribute to the existing theories of governance and education, theories such as system theory, constructivism, postmodernism, and critical theory. Moreover, further research is needed to improve the proposed framework to establish deeper comprehension of the practical work on data-driven policy and action.

We evaluated the Saudi government's performance in planning and governance of education in the country during the pandemic against well-known good governance international standards (ESCAP, 2009) and evidenced high achievements in terms of its participatory approach (Ministry of Education, 2020), speed of responsiveness (Algaissi et al., 2020; Hassounah et al., 2020), transparency (Saudi National eLearning Center (NELC), 2020), accountability, inclusivity, equity, and social sustainability (Saudi Ministry of Education, 2020b). This could be seen in the outcomes, e.g., low infection rates and deaths (Worldometer, 2022), sustained attendance and performance in schools (Official Twitter Account of Saudi Press Agency, 2022; Saudi Press Agency, 2022), collaborations and commendations from international institutions for effective education delivery in Saudi Arabia during the pandemic (Saudi National eLearning Center (NELC), 2020; Reimers and Opertti, 2021; Saudi Press Agency, 2021), etc. As we have discussed in the Parameter analysis and findings section in detail, the government provided all the citizens access to online learning through free TV lectures (Ministry of Education, 2022a; Saudi Ministry of Education, 2022) and other incentives such as financial subsidies for internet access (Saudi Ministry of Education, 2020a) to maintain equity and social sustainability. The government designed several initiatives and programs to nurture responsibility, honesty, and positive behavior among students (Madrastati, 2020; Saudi Ministry of Education, 2020c, 2021b). The government also made efforts to create understanding, empathy, compassion, appreciation, and harmony among teachers, families, and other stakeholders (Saudi Ministry of Education, 2020b). No issues were discovered from the Twitter data by our software concerning the lack of equity in the provision of access to online educational resources, lack of food, security, or other basic needs of people. The provision of free COVID-19 treatment and vaccination allowed people to focus on education without worrying about food, healthcare, and vaccinations (Unified National Platform GOV.SA, 2022).

The government implemented effective pandemic measures that allowed successful transitions between online and in-class learning (Saudi National eLearning Center (NELC), 2020) without affecting the educational goals while keeping people safe through physical distancing and other pandemic measures (Alomari et al., 2021a). Saudi Arabia was ranked number 6 in the world in March 2022 in “Covid Resilience” by Bloomberg (Lew et al., 2022) (earlier it was ranked number 3 in the world). They considered, in this ranking, various factors related to “Reopening Progress,” “Covid Status,” and “Quality of Life”. For evaluating the overall Saudi management of the COVID-19 pandemic that helped with the transitions between online and in-class learning, we also considered the Eurasia Group's methodology for evaluating and ranking international pandemic responses; the methodology focused on three key dimensions: healthcare management, financial policy response, and political response (Bremmer, 2020; Alam et al., 2021). The Saudi government has acted well in all these three areas (see, for example, Alomari et al., 2021a). The government was efficient and effective in healthcare management during the pandemic (Leaders Mena Magazine, 2020; AlFattani et al., 2021). For example, the government implemented quick measures to contain and control the spread of the virus (e.g., quarantines, physical distancing, etc.), provided free treatment to all citizens, residents, and even to the residents who violated the regulations of residency (Ministry of Health, 2020; Saudi Press Agency, 2020), and launched a range of telehealth applications to suit different public segments (Algaissi et al., 2020; Alghamdi et al., 2020; Hassounah et al., 2020). Saudi government acted well in maintaining physical distancing in daily life (Alomari et al., 2021a) as well as a successful Hajj season that involves a large gathering of people with zero COVID-19 cases in 2021 (El-Ziq, 2021; Saudi Gazette, 2021). There was sufficient availability of and operational access to the COVID-19 vaccines in the country (Bell, 2021). The government provided various financial initiatives and incentives to support businesses during the COVID-19 pandemic in order to mitigate the negative financial and economic impacts of the pandemic (Saudi Ministry of Investment, 2020). For example, a financial incentive of nine billion Saudi riyals was provided to the private sector employees in the country to contribute toward 60% of their salaries by the government (Saudi Ministry of Finance, 2020; WAS Saudi Press Agency, 2020). Another financial package of 6 billion Saudi riyals was provided to allow delaying mortgage payments and rescheduling loans (Saudi Ministry of Investment, 2020). Moreover, the government had allocated a sum of 47 billion riyals to the healthcare sector to improve the sector's readiness (Saudi Ministry of Investment, 2020).

The benefits and limitations of social and digital media-based research and practice have been a subject of much discussions such as data ownership (Montgomery, 2017), difficulty in enforcing ethical compliance (Sparks et al., 2016), people's safety (Huang et al., 2016; Facca et al., 2020), privacy (Al-Dhubhani et al., 2018; Editorial, 2020), security (Terzi et al., 2016; Carter, 2019), and more. However, non-digital methods for policy and action research are too limited, slow, and practical. Alam et al. (2021) provide in their study a concise summary of the limitations and solutions of digital data in research and the references therein for further details. There are rising and important concerns related to green, explainable, AI (Yigitcanlar et al., 2021b), and responsible innovation (Yigitcanlar et al., 2021a), and these must be addressed should we expect to reap the benefits of data-driven (sustainable) smartization. Continued research and awareness of the limitations of digital, data-driven, and AI methods will lead to improvements in methods and reduction of risks and we will continue on this mission in the future.

We live in a complex world comprising many complex systems. New approaches to governance are needed to tackle the global challenges. Smartization could deliver a sustainable future for us due to its underpinning of data-driven analytics-based policy and action for social, environmental, and economic sustainability. This study sought to provide a case study on sustainable participatory governance using a data-driven discovery of parameters for planning physical and digital learning in Saudi Arabia during the ongoing pandemic. We developed a software tool comprising a complete machine learning pipeline and used a dataset comprising around 2 million tweets collected from October 2020 to December 2021 and discovered fourteen governance parameters grouped into four governance macro-parameters.

The results clearly demonstrated the possibility and benefits of our sustainable participatory planning and governance approach, allowing the discovery and grasp of important dimensions of the education sector in Saudi Arabia; the complexity of the domain, stakeholders, and decision making; and the challenges and opportunities offered by the evolving national and international landscape. The tool allows governments to have a participatory approach toward governance and improve policy and actions perpetually through feedback from all parties.

We have built significant capacity in data-driven urban computing research such as our research on improving machine learning-based methods, and application of these methods, and the use of emerging technologies in smart societies and several urban sectors (Mehmood et al., 2017b, 2020; for example, see Alam et al., 2017; Alyahya et al., 2020; Arfat et al., 2020; Ahmad et al., 2022; Alahmari et al., 2022; Janbi et al., 2022). We will continue to build this capacity further with our mission of contributing to the international efforts on developing smarter sustainable societies.

The datasets presented in this article are not readily available because the data is collected from Twitter and hence we abide by the associated restrictions. Requests to access the datasets should be directed to RMehmood@kau.edu.sa.

SA and RM conceived, developed, analyzed, and validated the study. SA developed the software. SA and RM prepared the initial draft, and RM and IK reviewed and edited it. RM and IK supervised funds and resources and contributed to the article editing. All authors contributed to the article and approved the submitted version.

This project was funded by the Deanship of Scientific Research (DSR) at King Abdulaziz University, Jeddah, under grant number RG-6-611-40. The work carried out in this study was supported by the HPC Center at the King Abdulaziz University (KAU).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.