A group decision-making model for architectural programming in megaprojects

Megaprojects play a crucial role globally (Flyvbjerg, 2014), serving as public works that provide fundamental services for national development (Tan-Mullins et al., 2017), urban construction (Li et al., 2018), and public life (de Faria et al., 2017). However, their success often faces significant challenges due to characteristics such as large investment scale (Flyvbjerg, 2014), extended construction periods (Ma et al., 2017), numerous unknown risks (Kardes et al., 2013), complex functional organization (He et al., 2015), diverse user groups (Zhou et al., 2021), varied project requirements (Szyliowicz and Goetz, 1995), high uncertainty conditions (Sanderson, 2012), and involvement of complex stakeholder interests (Xue et al., 2023).

Architectural programming, also known as project briefing, serves as the initial stage before project construction, undertaking tasks such as defining project goals, positioning, requirements, risks, and resources (Hershberger, 2015). It elucidates the values of clients, users, and the public (Pena and Parshall, 2012), influencing the entire project lifecycle (Gray and Larson, 2011). Architectural programming primarily involves the three steps of collecting, analyzing, and synthesizing (Kelly and Duerk, 2002) to assist in making correct decisions during the design, construction, and operation phases (Blyth and Worthington, 2010). In megaprojects, traditional architectural programming faces challenges in establishing dynamic and shared understanding and commitments among various stakeholders due to their differences and preferences, posing risks to project success (Bouchlaghem et al., 2000).

Group Decision-Making (GDM), as a decision-making method that aggregates individual interests into collective benefits (Black, 1958; Kacprzyk and Fedrizzi, 1990; Roubens, 1997), has been introduced into the research domain of megaprojects (Luo et al., 2011). It aims to coordinate conflicting interests among parties and alleviate the issue of centralized decision-making in the decision process, achieving fairer and more scientific decision outcomes. Currently, numerous group decision-making models have been successfully applied to megaprojects, including artificial neural networks (Khosrowshahi, 1999), fuzzy preference relations (Chiclana et al., 2007), least squares path modeling (Liu et al., 2015), machine learning (Fenza et al., 2021), providing new avenues for multi-agent decision participation.

Previous studies have primarily focused on various aspects of architectural programming in megaprojects, including the clarity of project briefs (Vahabi et al., 2022), stakeholder involvement (Luo et al., 2011), project management efficiency (Jelicic et al., 2023), and optimization of project portfolio decisions (Hashemizadeh and Ju, 2019). However, research on effectively addressing multi-stakeholder conflicts, improving decision quality, and enhancing integration among stakeholders remains relatively limited. Building upon this foundation, this study proposes an architectural programming methodological system (APMS) for megaprojects based on group decision-making model, aiming to address the challenges in multi-stakeholder decision-making in current architectural programming for megaprojects. By adopting a framework of group consensus and establishing optimal objective solutions based on multi-stakeholder preference relationships, APMS aims to enhance efficiency and decision quality in group decision-making. This study elaborates on the literature review, conceptual framework, and application evaluation in sequential order, while also discussing the limitations and contributions of the research.

This section discusses some previous attempts made to address the decision-making process of architectural programming in megaprojects. Table 1 summarizes some studies conducted previously to research architectural programming in megaprojects.

The existing literature on architectural programming in megaprojects covers multiple research topics. Firstly, researchers have extensively focused on the critical success factors in architectural programming for megaprojects, identifying factors such as communication, briefing documents, client intentions, project scope, and leadership as crucial for project success (Yu et al., 2006; Tang et al., 2013, 2015; Yu and Shen, 2015; Surlan et al., 2016; Xiang et al., 2021; Lee et al., 2022; Opoku et al., 2024). Secondly, researchers have developed architectural programming software systems for megaprojects using computer-aided tools. These systems assist in handling client requirements, enhancing design standards, and improving the efficiency and quality of architectural programming (Kamara and Anumba, 2001; Hansen and Vanegas, 2003; Luo et al., 2010; Shen et al., 2013). Thirdly, establishing theoretical frameworks for architectural programming in megaprojects is also an important area of research, including collaborative work plans, automated generation frameworks, programming evaluation frameworks, and scheme programming frameworks (Chung et al., 2009; Khosrowshahi, 2015; Milovanovic et al., 2023; Medic et al., 2024). Fourthly, researchers have promoted decision efficiency in architectural programming by clarifying the characteristics and objectives of megaprojects, thereby enhancing the value proposition of projects (Bogenstätter, 2000; Kalayci and Ozdemir, 2021). Fifthly, researchers have summarized the challenges and opportunities faced in architectural programming for megaprojects and proposed corresponding improvement suggestions (Deng and Poon, 2013; Park-Lee and Person, 2018). Sixthly, comparative studies of architectural programming methods between different countries and methodologies have also attracted attention. This cross-cultural and interdisciplinary perspective has facilitated a comprehensive understanding of architectural programming (Yu et al., 2008; Al-Shalche and Al-Dabbagh, 2022). Lastly, researchers have discussed the scope and definition of megaprojects, and the ethical issues in architectural programming (Gibson and Gebken, 2003; Fellows et al., 2004).

Various research methods have been employed in the existing literature on architectural programming for megaprojects, with researchers selecting different methods based on their research purposes and the characteristics of the problems being addressed. Firstly, many studies have utilized qualitative research methods such as literature reviews, case analyses, and expert interviews to gain insights into the key objectives, challenges, and opportunities in architectural programming for megaprojects (Fellows et al., 2004; Chung et al., 2009; Tang et al., 2013; Xiang et al., 2021; Abe et al., 2023). These qualitative research methods draw lessons from practical experience and professional knowledge, providing valuable theoretical support and practical guidance for architectural programming. Secondly, some studies have employed quantitative research methods, including questionnaire surveys, mathematical models, and factor analyses, to quantify and rank the key factors in architectural programming, thereby providing scientific evidence and data support for decision-making (Kamara and Anumba, 2001; Yu et al., 2006; Shen et al., 2013; Yu and Shen, 2015; Tang et al., 2015). These quantitative research methods, through the collection and analysis of large-scale data, reveal the relationships and degrees of influence between different factors, providing important references for project management and decision-making. Additionally, some studies have employed mixed research methods, integrating the strengths of qualitative and quantitative research to comprehensively explore the complexity and diversity of architectural programming for megaprojects from multiple perspectives (Khosrowshahi, 2015; Surlan et al., 2016; Lee et al., 2022; Milovanovic et al., 2023). These mixed research methods make full use of the complementarity between qualitative and quantitative data, enhancing the credibility and persuasiveness of the research.

In conclusion, significant progress has been made in the field of architectural programming for megaprojects, with in-depth discussions and research on critical success factors, computer-aided tools, theoretical frameworks, characteristics and objectives, challenges and opportunities, cross-national comparisons, and ethical issues. However, there are still some research gaps that need to be addressed. Firstly, although there have been some studies focused on the architectural programming process for megaprojects, there is relatively less discussion on how to effectively optimize the group decision-making process. Particularly, there is a lack of systematic research on the design, application, and evaluation of group decision-making models. Secondly, research on the roles and influences of stakeholders in group decision-making processes is also relatively limited in the existing literature. In architectural programming for megaprojects, stakeholders often come from different fields and levels, with varying interests and perspectives. Effectively integrating stakeholders’ opinions and requirements is thus an important challenge.

Optimizing the decision-making process plays a crucial role in architectural programming for megaprojects. To address this challenge, methods based on group decision-making theory are commonly employed.

Luo et al. (2011) developed a Group Decision Support System (GDSS) based on the Value Management (VM) implementation approach to tackle the complexity in architectural programming for megaprojects. The system aims to meet the requirements of a computer-supported collaborative working environment, based on the underlying logic of the VM method. It allows different clients to define requirements through functional ideas and evaluate and strengthen them based on Functional Performance Specifications (FPS) for further development in the design phase. The strength of GDSS lies in its knowledge-based approach, enabling users to manage previous projects through retrieval, facilitating the participation and interaction of briefing teams, thus shortening the time for VM formulation. Despite the potential of this system in enhancing VM performance, its limitations include a lack of comprehensive consideration and in-depth analysis of the needs of different stakeholders and the potential issue of information overload when dealing with complex decisions.

Tu and Chen (2015) constructed a multi-agent information platform based on group decision-making to address the selection and preference issues of various interest groups in megaprojects, promoting quantitative research in the early stages of project programming. The platform achieved comprehensive decision-making by considering decision-makers’ preferences, weights, and various aspects of the decision-making process. Although the platform brought a degree of quantification and multi-stakeholder participation to project programming, it still has some limitations. Firstly, the modeling of decision-makers’ preferences may suffer from subjectivity and uncertainty, affecting the accuracy and reliability of the model. Secondly, the survey method of the platform may be constrained by sample selection and data collection methods, resulting in bias and uncertainty in decision results. Additionally, the platform may face challenges in reducing decision efficiency when dealing with complex decisions.

The primary objective of this study is to provide a scientific and fair decision-making platform for megaprojects through APMS, to assess and determine the decision preferences and outcomes of various stakeholders involved in megaproject architectural programming. The proposed conceptual framework is illustrated in Figure 1, with selection criteria derived from a review of previous literature on megaproject architectural programming and interviews with experts.

In architectural programming, early involvement of multiple stakeholders in decision-making is one of the key factors determining the success or failure of megaprojects. Based on relevant national standards and literature, this study categorizes the decision-making subjects of APMS into 5 key groups: the client, the expert, the user, the government, and the public. As shown in Table 2, different decision-making subjects have varying degrees and methods of influence on megaprojects.

Among these, the client, as the project sponsor and ultimate beneficiary, primarily focuses on the project’s economic benefits, the effectiveness of goal achievement, and overall investment returns. Their decision preferences typically prioritize successful project delivery and economic benefits (Fellows et al., 2004; Khosrowshahi, 2011). Experts possess knowledge and experience in specific fields, providing essential insights into the technical feasibility and design innovation of the project. Their decision preferences usually focus on the technical and implementation feasibility of the project (Luck and McDonnell, 2006; Yu et al., 2006). Users are sensitive to the project’s usability and environmental perception. Their decision preferences are often closely related to factors such as the functionality, practicality, and environmental impact of the project (Edwards, 2006; Kamara and Anumba, 2000; Shen et al., 2013). Governments, as regulators and policymakers, are more concerned with the project’s compliance, social welfare, and public benefits. Their decision preferences involve the social benefits and regulatory governance of the project (Milovanovic et al., 2023; Park-Lee, 2020). The public, as part of society, focuses on the project’s impact on the community, cultural value, and level of public participation. Their decision preferences often focus on the social sustainability, cultural preservation, and community integration of the project (Stafford, 2013; Xue et al., 2020, 2021).

In the architectural programming of megaprojects, decision objects refer to comprehensive factors related to the project, and a comprehensive understanding of these factors is crucial for scientific decision-making. Currently, several authoritative theories have proposed research and classification of decision objects in architectural programming (Pena and Parshall, 2012; Hershberger, 2015). Based on literature review and expert experience, this study divides the decision object into seven dimensions: environment, humanity, society, function, form, economy and time, as shown in Table 3.

Among them, the environmental dimension emphasizes the correlation and impact of the project with the surrounding environment, including the site characteristics such as topography and soil, the meteorological conditions such as temperature and humidity, and the urban background such as planning guidelines and surrounding architectural styles.

The humanistic dimension emphasizes the impact of the project on human subjective feelings, including the physical sensations such as vision and hearing, the physiological needs such as ventilation and lighting, and the psychological effects such as emotion and comfort.

The social dimension focuses on the project’s contribution to society and societal expectations, including the cultural characteristics such as architectural styles and artistic elements, the legal compliance such as building codes and environmental regulations, and the social impact such as public recognition and acceptance.

The functional dimension focuses on the project’s practicality, execution effectiveness, and alignment with expected functional goals, including the behavioral requirements such as building use and functional zoning, the spatial environments such as design layout and human factors scale, and the advanced technology such as smart systems and sustainable technologies.

The form dimension emphasizes the project’s appearance and aesthetics, including the overall style such as appearance coordination and uniqueness, the material sustainability and the structural stability, and the aesthetic effects such as color matching and artistic elements.

The economic dimension focuses on the project’s economic feasibility and long-term maintenance, including the initial investment and cost planning, the long-term operating costs such as maintenance and energy consumption, and the post-construction maintenance costs and the economic benefits during the project’s use phase.

The time dimension focuses on the project’s continuity and evolution over time, including the historical background and cultural accumulation, the social needs and environmental adaptability, and the future development potential and social development impact.

Given the multitude of factors involved in architectural programming for megaprojects, the classification of decision objects needs to consider the interaction of multiple dimensions to achieve hierarchical output of final decision information. Building upon the “problem-structuring method” (Pena and Parshall, 2012), this study divides the decision method into five key aspects: objectives, concepts, facts, requirements, and issues. As shown in Table 4, the decision method decomposes the establishment of decision objects into different key elements, thus more effectively handling complex information in the programming process.

Among these, “objectives” aim to provide clear direction for subsequent decisions by defining the project’s vision, mission, and overall goals. “Facts” provide ample factual basis for subsequent decisions through systematic data collection and analysis. “Concepts” propose different design concepts and test their feasibility to find the optimal solution. “Requirements” involve decision-making regarding various project needs, including spatial requirements, construction quality standards, budget, and time, ensuring the reasonability and adjustability of each requirement. “Issues” aim to clarify and identify potential problems, providing a basis for possible adjustments and optimizations in the decision-making process.

This study employs the entropy weight method and fuzzy Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) to address the multidimensional, multi-stakeholder decision-making issues in architectural programming for megaprojects (Chen, 2019; de Boer et al., 2005; Palczewski and Salabun, 2019). The reasons for selecting these algorithms are as follows: Firstly, the entropy weight method can effectively handle the correlation between multiple dimensions and calculate weights, ensuring reasonable considerations and balances among dimensions in the decision-making process. Secondly, fuzzy TOPSIS can evaluate solutions from the perspectives of different groups, integrating the opinions and requirements of various groups to derive more comprehensive and representative decision results. Additionally, it can effectively deal with fuzzy information by transforming fuzzy evaluations into specific numerical values through fuzzy set theory, enabling comprehensive evaluation and ranking. Finally, the entropy weight method and fuzzy TOPSIS can quantitatively convert experts’ subjective evaluations and experiences into computable values, making the decision-making process more scientific and objective (Boran et al., 2009; Kaya and Kahraman, 2011; Kumar et al., 2017). As shown in Table 5, symbols and parameters are defined:

• Step 1: Calculate the weighted decision matrix $Xij$ for decision-making subject $n$⁠.

Each decision-making subject $n$ needs to assess and score each decision object $j$ of the megaproject using decision method $i$⁠, resulting in the decision matrix $Xij$⁠:

Use the entropy weight method to calculate the entropy $Ei$ for each column of the decision matrix $Xij$ to measure the uncertainty of each decision method on decision objects:

Compute the weight $wi$ for each decision method $i$ on decision objects $j$⁠:

Normalize $w$ so that $∑i=1nwi=1$⁠. Finally, obtain the weighted decision matrix $Xij$ for each decision-making subject $n$⁠:

• Step 2: Normalize the weighted decision matrix $Xij$ for all decision-making subjects $n$⁠.

Calculate the normalized value $Rij$ for the decision matrix $Xij$ of each decision-making subject $n$⁠. The purpose of this step is to standardize the decision values of each decision-making subject involved in the project to make them comparable:

Where $j$ is the number of decision objects. The resulting $Rij$ represents the normalized values of decision object $j$ on different decision methods $i$ for all decision-making subjects $n$⁠.

• Step 3: Calculate the weighted decision matrix $Vij$ for all decision-making subjects $n$⁠.

Before obtaining the comprehensive decision matrix $Vij$ for all decision-making subjects $n$⁠, the Entropy method is similarly used to calculate the weight of each decision-making subject $n$ for the decision outcome. This method can allocate weights to different decision objects while considering the decision-making subject’s situations. Specifically, for each column of decision objects $j$ in the decision matrix $Rij$⁠, calculate its entropy $En$⁠:

$En$ measures the uncertainty of each decision-making subject on decision objects, and further, the weight $Wn$ for each decision-making subject can be calculated:

Through normalization, ensure the sum of all weights is 1. Finally, for each decision-making subject $n$⁠, calculate the weighted value of each $Vij$⁠:

• Step 4: Calculate the ideal and anti-ideal solutions.

After obtaining the weighted decision matrix $Vij$ for each decision-making subject $n$⁠, the ideal solution $Aj+$ and anti-ideal solution $Aj−$ for each decision object $j$ need to be calculated for subsequent optimal decision-making results. Specifically, the ideal solution takes the maximum value in each decision-making subject’s decision, while the anti-ideal solution takes the minimum value:

• Step 5: Calculate the distance from decision objects to the ideal and anti-ideal solutions.

Use Euclidean distance to calculate the distance of each decision object $j$ from the ideal and anti-ideal solutions, to measure the degree of proximity to the ideal solution and distance from the anti-ideal solution:

• Step 6: Calculate the comprehensive evaluation index.

Compute the comprehensive evaluation index $Ci$ for each decision object $j$⁠, representing the relative distance of decision object $j$ from the ideal solution $Aj+$ and anti-ideal solution $Aj−$⁠:

Finally, rank decision objects j from high to low according to the comprehensive evaluation index $Ci$ to obtain the optimal result for megaprojects architectural programming.

Through the index establishment and algorithm steps mentioned above, the APMS model is initially constructed in this study. As shown in Figure 2, the process design of APMS in the actual project programming process will be explained in detail below, including four main parts: preparation stage, information stage, decision stage and evaluation stage.

In the APMS, the preparation stage serves the purpose of information absorption and project understanding. This stage includes the following four steps: (1) Researching project background: Deeply understanding the project’s history, geography, social, and cultural aspects enable architects to have a comprehensive understanding of the project, providing direction for subsequent decisions. (2) Determining project types: Defining project types clearly to identify research directions, considering decision issues, and focal points for different types of projects. (3) Analyzing project characteristics: In-depth analysis of key project characteristics, identifying factors that need prioritization, laying the foundation for subsequent decisions. (4) Establishing project goals: Clearly defining decision objectives and desired outcomes to provide clear goals for defining decision issues.

The information stage aims to comprehensively understand decision-making subjects and decision objects, and to reabsorb and process information. It includes two modules: decision-making subject establishment and decision object establishment.

Decision-making subject model: (1) Comprehensive collection of information from various stakeholders. (2) Classifying decision-making subjects based on attributes such as experts, government, public, stakeholders, etc. (3) Selecting decision-making subjects based on principles such as information, responsibility, influence, etc. (4) Determining weight parameters for each decision-making subject for subsequent calculation of weighted decision matrix. (5) Determining the participation methods and timing for each decision-making subject to ensure full participation.

Decision object model: (1) Comprehensive collection of information related to programming, including qualitative, quantitative decision objects, and impact objects. (2) Classifying decision objects based on attributes such as qualitative, quantitative, impact, etc. (3) Adjusting the focus on decision objects based on different programming stages. (4) Classifying decision objects based on decision-making subjects’ focus. (5) Ranking decision objects by importance to ensure that major objects receive attention and efficiency is improved.

In the decision-making stage, decision-making subjects, decision objects, and decision methods constitute the decision matrix. With the support of group decision-making theory algorithms, decision-making subjects preferentially select decision objects in different dimensions of decision methods. The results are calculated using the entropy method and fuzzy TOPSIS algorithm, and preliminary decision conclusions are output. Subsequently, the design conditions and content are tabulated using computer modeling, resulting in a complete and logical decision report.

The evaluation stage is the process of rechecking the decision results after the decision-making stage. It includes two steps: validating decision results and decision feedback.

1. Validating Decision Results: Checking the consistency between decision outcomes and decision-making subjects. If the consistency is high, the final decision conclusion can be output; If the consistency is low, the survey results should be fed back to the primary stage, the relevant data should be corrected, and the correct results should be calculated again.

2. Decision Feedback: Evaluating the satisfaction of decision results and conducting a “group decision-making” process specifically for decision results, utilizing the results to retroact on the reasons to adjust decision activities. This includes decision-making subject feedback, feedback of statistical data from various stages, feedback on design effects, etc. Overall, the evaluation stage ensures that APMS considers and balances the needs of various decision-making subjects, forming a closed-loop system for the entire model.

This study selects the Shanghai Hongqiao Comprehensive Transportation Hub project as the actual application case of APMS. As one of the few megaprojects in China that has undergone comprehensive programming before construction, the project has undergone extensive verification and research over a long period. With its large scale, it has had profound impacts on Shanghai and surrounding cities, covering the needs of multiple stakeholders. Previous studies have mainly focused on describing and evaluating the planning, construction, and operation of the project, and its impact on urban development and socio-economics (Peng and Shen, 2016; Duan et al., 2021). However, this study focuses on the application of APMS, aiming to assess the needs of various stakeholders and compare them with the decision content of the early programming stage of the project to verify the effectiveness and reliability of APMS.

The Shanghai Hongqiao Comprehensive Transportation Hub was approved in 2005, commenced construction in 2006, and was put into operation in 2010, with a total investment of over 36 billion yuan. The project covers an area of over 1.3 million square meters, integrating railways, aviation, maglev, subways, light rails, buses, passenger stations, and taxis, forming a comprehensive transportation center (Figure 3).

This study collected evaluation data on the Shanghai Hongqiao Comprehensive Transportation Hub after its operation through a survey questionnaire. The questionnaire covered three aspects of the project’s architectural programming, including functional layout, transportation modes, and transfer modes. The questionnaire was distributed both online and on-site, with a total of 1,000 questionnaires distributed and 955 successfully collected. Among them, 915 questionnaires were deemed valid, resulting in an effective rate of over 90%. The participants were mainly concentrated in the age group of 25–40, accounting for 32% of the total participants. They were categorized into five types of decision-making subjects: government, clients, experts, users, and the general public, with proportions of 48: 79: 238: 209: 341. The distribution of these groups covered stakeholders at various stages of the project, ensuring the comprehensiveness and representativeness of the survey results (Figure 4). The statistical results of the questionnaire survey are presented in Table 6.

Table 7 lists the decision results calculated by APMS for the Shanghai Hongqiao Comprehensive Transportation Hub based on the questionnaire data.

Comparing the decision results obtained by APMS with the content of the project’s preliminary programming, the research findings are as follows:

1. Regarding the functional layout, the decision results of APMS are basically consistent with the preliminary programming results of the project. The current functional layout of the project is from west to east: high-speed railway station - maglev station - Terminal 2 - Terminal 1.

2. Regarding transportation modes, the decision results of APMS indicate that decision-making subjects tend to choose long-distance transportation modes, with preferences ranging from high to low: aviation, inter-city (high-speed rail and EMU), and long-distance (high-speed rail and EMU), which is consistent with the transportation mode ratio set in the early programming stage of the project.

3. Regarding urban transportation modes, the decision results of APMS indicate that decision-making subjects’ preferences for transportation modes are ranked as follows: subway, taxi, airport bus, bus, and self-driving. The project reduced the size of the parking lot in the preliminary programming stage to encourage users to prefer public transportation over self-driving, which is consistent with the decision results obtained from feedback after use. Additionally, the decision results of APMS indicate that setting up pedestrian walkways containing various types of commercial activities within the Hub is the optimal internal transportation mode, and users have a high acceptance of this mode.

4. Regarding transfer modes, there are some differences between the decision results of APMS and the project’s preliminary programming. Firstly, the decision-making subjects participating in the survey show a clear preference for shuttle buses as a transfer mode, while according to the project’s preliminary programming and current situation, shuttle buses have been discontinued after the subway lines were connected to Terminal 1, Terminal 2, and the high-speed railway station. Secondly, the preference of decision-making subjects for maglev as a transfer mode is relatively low, which differs from the project’s preliminary programming and current situation. The reason may be that the cost of this transfer mode is relatively high compared to others.

This study collected evaluation data on the functional layout, transportation modes, and transfer modes of the Shanghai Hongqiao Comprehensive Transportation Hub project from multiple decision-making subjects through empirical surveys. By using APMS to calculate these data, a series of decision results were obtained and compared with the content of the project’s preliminary programming. The results show that the decision results obtained by applying APMS in actual projects are basically consistent with the decision content of the project’s preliminary programming, and can effectively identify decision factors that can be optimized in the subsequent use phase.

APMS addresses the shortcomings of existing architectural programming theories and methods in measuring the involvement of multiple stakeholders in decision-making. Traditional methods struggle to accurately measure the impact of multiple stakeholders in the decision-making process, while APMS, by introducing group decision-making and complexity science, provides a more accurate architectural programming framework, making resource allocation in megaprojects more scientific.

Moreover, APMS not only promotes accuracy in architectural programming but also enhances transparency in the decision-making process. By establishing decision indicators and weights, APMS provides quantitative research tools for early-stage project programming, making the decision-making process more transparent and visible. This transparency helps stakeholders better understand the basis and process of decision-making, reduces information asymmetry, and increases the rationality and credibility of decisions.

APMS, by strategically establishing information matrices, can comprehensively understand the issues faced by megaprojects, thereby promoting the active participation of multiple stakeholders. By breaking away from the traditional architect-assisted owner decision-making model, APMS creates a more equal participation opportunity for stakeholders from various social, economic, and cultural aspects. This comprehensive engagement approach help achieve a balance of interests among multiple stakeholders in megaprojects, enhancing the overall value of the project.

Additionally, APMS can generate decision results more rapidly and efficiently in the early stages of projects, which accelerates project progress and promotes stakeholder integration. For post-occupancy evaluation of the project, APMS can continuously optimize and balance interests through the evaluation stage, helping to reduce design rework and resource waste, and better achieve project construction goals.

This study proposes an APMS framework based on group decision-making to better understand and improve the decision-making process in architectural programming for megaprojects. Existing literature primarily focuses on project briefing clarity, stakeholder participation, project management efficiency, and project decision optimization. However, these studies often overlook how to comprehensively integrate and balance the needs and opinions of various stakeholders during the decision-making process, which can lead to information asymmetry and interest imbalance issues.

Addressing this research gap, this study defines the decision-making subjects, decision objects, and decision methods within the APMS decision-making process, providing new theoretical support for complex decision-making in megaprojects. Specifically, the study introduces a new programming process that includes four key stages: preparation, information, decision, and evaluation. This process not only makes the entire decision-making procedure more orderly and systematic but also provides a clear guidance framework for group decision-making.

Through practical application and evaluation in the Shanghai Hongqiao Comprehensive Transport Hub project, this study validates the effectiveness and reliability of APMS in practice. The results indicate that, in terms of transportation modes, the decision-making results of APMS are consistent with the preliminary programming outcomes of the project. However, regarding transfer modes, the APMS decision-making results revealed certain discrepancies between the project's current status and the preliminary programming. The contributions of this study are as follows:

Improving Decision Accuracy and Transparency: APMS enhances decision accuracy through quantitative decision methods and increases stakeholder trust through a transparent decision-making process.

Effectively Integrating Stakeholder Needs: APMS effectively collects and analyzes information from various stakeholders, ensuring comprehensive consideration and balance of all parties' needs during the decision-making process.

Despite these achievements, some limitations were identified in practical applications. Firstly, the empirical research was conducted after the project was completed and operational, thus failing to intervene in the early stages of the project. Secondly, the limited sample size of the survey might affect the generalizability of the research. Future research can explore the following areas: firstly, intervening in the early stages of actual projects to form a closed-loop study of project construction. Secondly, establishing AI-based decision algorithms to make decisions using extensive project data. Lastly, developing APMS using computer software to better serve the construction of megaprojects.

All authors would like to thank the respondents who participated in this study, as well as the editors and reviewers for their valuable advice on this study.

The supplementary material for this article can be found online ( supplementary material 1: Questionnaire and  supplementary material 2: Research data for practical projects).