Residential energy retrofit decision-making is hindered by an expertise gap, where homeowners lack the technical literacy required to conduct structured, expert-level building energy assessments, thereby limiting proactive energy retrofit planning. To bridge this expertise gap, this study reports a domain-specific large language model (LLM) that provides optimal retrofit recommendations based solely on natural-language descriptions of basic dwelling characteristics from homeowners. The model is fine-tuned on a large corpus grounded in 536,416 U.S. residential building prototypes, integrating physics-based energy simulations and techno-economic calculations. Nine major retrofit categories are evaluated, including envelope upgrades, HVAC systems, and renewable energy installations. The model uses parameter-efficient Low-Rank Adaptation (LoRA) on LLM architecture to map homeowner-accessible descriptions (e.g., building age, size, and location) to optimal performance outcomes. The outputs against physics-grounded baselines demonstrate that this domain-specific LLM identifies the optimal retrofit for CO₂ reduction within its top three recommendations in 98.9% of cases and the shortest discounted payback period in 93.3% of cases. The fine-tuning significantly improved the accuracy of performance predictions, yielding an order-of-magnitude reduction in prediction error for CO₂ reduction, and multi-fold reductions for net site energy reduction and retrofit cost. Furthermore, the model exhibits high stability under incomplete input conditions, maintaining reliable selection accuracy even when up to 40% of non-core building parameters are masked. The results indicate that the proposed domain-specific LLM can effectively synthesize complex building physics and techno-economic trade-offs with an accessible human-machine interface, reducing subjective uncertainty and providing an informed decision-support tool for residential decarbonization initiatives.

Residential energy retrofit initiation is often stalled by an expertise gap, where homeowners lack the technical literacy required for structured building energy assessments and are thereby trapped in low-information environments with fragmented sources. To bridge this gap, this study reports a domain-specific large language model (LLM) designed to catalyze informed decision-making based solely on homeowner-accessible, natural-language descriptions, e.g., building age, size, and location. The model is created using the parameter-efficient low-rank adaption (LoRA) fine-tuning approach on a massive corpus grounded in physics-based energy simulations and techno-economic calculations from 536,416 U.S. residential building prototypes. Nine major retrofit categories are evaluated, including envelope upgrades, HVAC systems, and renewable energy installations. Validations against physics-grounded benchmarks show that the LLM consistently identifies high-quality retrofit options, achieving top-3 hit rates of 98.9

Exposure to teamwork and collaborative projects in engineering education is crucial for preparing students for engineering jobs. Several universities are adopting project team-based learning (PBL) to deliver work-ready graduates required in technically complex interorganizational project environments. However, the use and levels of adoption of project management techniques and tools (e.g.,charters) and the multilevel nature of team science for engineering education and workforce development are not well-investigated in the literature. This study investigates whether the level of perceived compliance with using a project charter at individual and team levels throughout project delivery enhances team performance (i.e.,group potency, team viability, and team cohesion) in engineering education settings, such as classroom and cocurricular projects. We examined relationships using multilevel modeling (MLM), which means examining teams at both team member (individuals) and team levels (groups). We utilized qualitative insights to guide model specification and interpretation. The findings revealed that systematic use of project charters, assessed through perceived compliance, was linked to significant improvements in group potency, team viability, and team cohesion. Compliance effects were observed at both levels, with generally comparable contributions to group potency, team viability, and team cohesion. Furthermore, the link between member compliance and performance tended to be weaker in teams with high performance. Our MLM analyses also revealed that classroom teams scored lower at baseline on group potency, team viability, and team cohesion compared with cocurricular project teams. Future studies should delve deeper into these differences by examining factors such as work environment, modality of education (e.g.,classroom versus cocurricular, or for pay versus for a grade or experience), team composition (e.g.,level of education, experience, and skill heterogeneity), project nature, and prior relationships among team members.

The development of teamwork skills in project management is becoming increasingly important as the future of work requires increasingly complex team organization and coordination. Professionals of highly technical backgrounds face challenges in working in collaborative environments involving communication, management, conflict resolution, and challenging behaviors. To help improve construction engineering students' collaborative skill development in and via project team assignments, this paper presents an iterative development process of dynamic project team interventions. An intervention module was designed after a comprehensive literature review, which was later tested in a pilot study and revised as the final version. The module focuses on planning, management, coordination, conflict resolution, and communication aspects of project teamwork, and includes a dynamic charter, a training module, and reminders driven by teams' use of the project charter. Insights into training and intervention modules for the future of work in higher education are discussed, and directions for future research are presented.

The construction industry relies on temporary alliances for project completion, emphasizing effective coordination and communication. Line of privity (LoP) is a key concept in construction projects, representing the contractual relationship between parties and playing a vital role in inter-team coordination. This study investigates the impact of LoP on team communication and productivity within a transaction cost framework, using mixed methods with social network analysis. The main hypothesis posits that contractual and organizational ties, measured by LoP and distance, enhance project team productivity through improved communication. Data were collected from a healthcare project using mixed methods throughout project delivery. Analysis reveals the importance of direct communication, although excessive focus may hinder performance due to fewer parties involved or information overload. Increased organizational distance negatively affects completion rates, while frequent long-distance communication promotes understanding and coordination among distributed teams. A direct correlation between the ratio of LoP communications and project performance was observed. This paper provides insights for construction project management, emphasizing the significance of direct communication, effective management of organizational distance, fostering long-distance communication for distributed teams, and understanding the impact of LoP communications on project performance. These findings contribute to efficient coordination, reduced idle time, and minimized rework.

Most architecture, engineering, and construction (AEC) projects experienced disruptions, changes, or cancellations due to the COVID-19 pandemic. To compensate for the adversarial effects of this unexpected situation, some project owners stopped and restarted their projects after evaluating their scope and timeline, which caused changes in role assignments and communication structures. However, there is a knowledge gap regarding how communication network structures changed at the macro level with the involvement of new individuals at the micro level for the projects paused and restarted. Thus, this study examines the changes in the individuals occupying key roles and the impact of their egocentric networks on the project networks considering performance values. The research team collected archival, e-mail exchange, and survey data from an AEC project during the construction phase and conducted social network analysis. The study findings demonstrated that network topologies and collaboration patterns were different before and after the involvement of new project members. Especially, the new project manager acting as a network bridge prevented the formation of subgroup structure and contributed to the existence of the core-periphery structure. The results add to the body of knowledge by providing empirical observation on the macro-level effects of micro-level changes in AEC teams.

Collaboration is challenging for complex project teams that have many subteams with diverse and complementary skills that are working toward shared goals. Knowledge transfer among subteams (i.e., project-level knowledge transfer) enhances integrative collaboration and results in behavioral changes within subteams (i.e., group-level knowledge transfer). However, there is a lack of quantitative evidence about how such project-level knowledge transfer influences group-level knowledge transfer behaviors. This study examined the change in knowledge transfer behaviors along with project progress, using the energy-efficiency subteam as an example. To achieve the goal, the authors collected longitudinal email exchanges and archival project data during the delivery of a complex construction project in the state of Michigan. Social network analysis and social influence models were utilized to analyze the change of knowledge-transfer networks over time. The results confirmed that exposure to project-level knowledge transfer positively predicts the subsequent group-level knowledge transfer behaviors during project delivery, measured by eigenvector centrality. The findings provide quantitative evidence explaining the importance of knowledge transfer behaviors in project communication: project-level knowledge transfers change members' attitudes and behaviors and subsequently improve individuals' influences in their respective project subteams. In addition, opinion leaders demonstrate a certain extent of resistance to the change.

Successful execution of a project depends on efficient collaborations among a diverse set of actors and organizations. Establishing efficient collaboration networks is vital to channeling necessary expertise to achieve desirable outcomes. Network interventions describe the process of using network-supported strategies to enable knowledge sharing and coordination critical for network performance. However, there is a lack of intervention studies in the architectural, engineering, and construction (AEC) domain. To fill this gap in the literature, this study used social network analysis (SNA) and mixed methods to develop an intervention framework for AEC teams. The research team examined the performance and communication trends of an AEC project team during project delivery in light of the literature to inform network interventions. After vetting the developed strategies through interviews with select project leaders, interventions were presented to the whole project team for further verification and feedback. The results contribute to the body of knowledge by providing an intervention framework for AEC teams and implications for practitioners in selecting and assigning experts in key roles and positions in projects as well as managing team communications and networks in a dynamic manner during project delivery for improved team and project performance.

This study explored changes in communication patterns within construction project teams during times of disruption (i.e., the COVID-19 pandemic), focusing on transaction costs associated with communication and coordination through individual-level project team interactions. Nodes within the communication networks were classified based on their roles (owner, general contractor, designer) and were further categorized into three tiers according to their significance, decision-making authority, and involvement in the project. By utilizing social network analysis (SNA) and hierarchical linear modeling (HLM), the study discerned considerable shifts in network dynamics. Designers and owners in tier 2 showed a decline in network centrality and importance, as evidenced by reduced eigenvector centrality and increased inverse of closeness centrality. Conversely, the owner and general contractor in tier 1 either maintained or slightly enhanced their network importance. Local transitivity, indicating the propensity for nodes to form clusters, fluctuated across roles and tiers. The pandemic typically decreased the authority score, implying diminished perceived credibility of nodes. The elevated communication load on tier 1, mainly including project managers, exerted pressure on their capacity limits. The disruptive period and the potential impacts of COVID-19 marked an irrevocable shift in global communication, not just in the construction industry, necessitating role and tier-specific strategies.

Information bottlenecks are a central issue in architecture, engineering, and construction (AEC) projects. It is crucial to detect their occurrence to mitigate their impacts. Existing approaches to detect information bottlenecks use metrics that require estimating the quantity of information published, which is a challenging task in practice. In this paper, building on the literature and recent developments in automation, we introduce a revised model that uses unresolved activities during project delivery and frequency of information publication actions to detect information bottlenecks. We validated our model by applying it to a case study. We compared bottlenecks detected by our model to ground truth obtained in the case study independently using observational data and external verification. We further study how project activities requiring different levels of interactions among project participants with different roles (e.g., designer, contractor) affect our model’s performance. Results support our model and show that it is possible to attain comparable performance by considering only high-interdependency activities. Our research enables teams and project managers to keep a focus on issues that require intense communication among different experts and organizations to avoid bottleneck. Our contribution to the body of knowledge includes a revised model that detects information bottlenecks during project delivery and insights into integrative project management where the focus is on both project activities and collaboration needs across roles to streamline the whole process.

Keeping track of issues and their documentation in architecture, engineering, and construction (AEC) projects demand significant amounts of time, budget, and effort. While various types of documents and software aid coordination in AEC projects, project team meeting minutes, developed as a follow-up to periodic project team meetings, continue to be the most common and prominent type of documentation across project types for recording team communications, tasks, and assignments. Presently, due to its dynamic nature, identifying unique project issues and tracking their progress from meeting minutes is a manual process that is time-consuming and susceptible to error. This study aims to automate the identification of project issues and track resolution timelines using project team meeting minute documents via the Jaccard similarity method. In this study, over 50 AEC project team meeting minutes documents of varying formats from three different projects of various sizes were collected, automatically converted, and coded to train the Jaccard similarity model for detecting new and continuing issues. Accuracy, precision, recall, and F1 parameters were tested, and the accuracy rates of 81.86%-94.18% were obtained. The study provides the groundwork to automate the analysis of issue complexity, detection of bottlenecks, and analysis of expertise assignments for issue resolution.

This study examines the influence of project managers’ communication behaviors in integrated project delivery (IPD) projects. This study also considers the impact of team-level information sharing on team-level outcomes (i.e., goal alignment, decision quality, process commitment, and project quality). Results indicate that team information sharing was perceived as beneficial, but the effects of project manager communication behaviors were inconsistent. Interaction analyses indicated that project managers’ communication behaviors hindered team functioning if teams were already engaging in adequate levels of information sharing. This manuscript contributes to the project management corpus by explicating an integrative model whereby project manager and team-level information-sharing behaviors are modeled simultaneously.

Green architecture, engineering, and construction (AEC) projects are deemed more complicated compared with their traditional counterparts. They have been reported to have relatively more frequent delays and higher cost growth. The AEC literature has repeatedly highlighted the importance of leadership for sustainable design and construction and the need for team integration for optimal outcomes. However, studies exploring leadership in the context of green AEC projects are limited. Transformational leadership is a highly regarded theory that is known for its potential to effectively facilitate team integration. This empirical study tests the role of shared transformational leadership received by team members in improving team performance through team integration for green AEC project teams. We collected data from 103 individuals working in nine Leadership in Energy and Environmental Design (LEED) projects in the US. We analyzed the data using structural equation modeling (SEM). Our findings suggest that transformational leadership received by team members is positively related with perceived team performance, and this relationship is mediated through perceived team integration. The study is a pioneer in examining leadership for sustainability and mediation effects of team integration in the AEC literature. The outcomes urge for a change in green AEC project teams to think beyond project delivery systems and practices and incorporate transformational leadership across organizational and communicational tiers of project teams through team procurement, project management, and skill development and training.

Crowdsourcing has emerged as a popular approach for collecting annotated data to train supervised machine learning models. However, annotator bias can lead to defective annotations. Though there are a few works investigating individual annotator bias, the group effects in annotators are largely overlooked. In this work, we reveal that annotators within the same demographic group tend to show consistent group bias in annotation tasks and thus we conduct an initial study on annotator group bias. We first empirically verify the existence of annotator group bias in various real-world crowdsourcing datasets. Then, we develop a novel probabilistic graphical framework GroupAnno to capture annotator group bias with an extended Expectation Maximization (EM) algorithm. We conduct experiments on both synthetic and real-world datasets. Experimental results demonstrate the effectiveness of our model in modeling annotator group bias in label aggregation and model learning over competitive baselines.

The project-based nature of the construction industry requires organizations to establish temporary alliances to complete projects within required limitations (e.g., time, cost, and quality). While projects endure many risks during delivery, times of disruption (ToD), such as economic crises, extreme weather, and epidemics can create greater strains on teams, specifically hindering integration elements critical to success of project teams. Motivated by the need to improve team integration in inter-organizational and interdisciplinary construction project teams, in the light of social network theory, this study poses to identify problems regarding transaction costs (TC) and knowledge transfers (KT) during ToD: breakout of the COVID-19 pandemic. To achieve this aim, the researchers conducted and analyzed interviews with core members of a complex project team involving over 85 organizations. The results provide insights to common communication problems in complex project teams leading to suboptimal team and project performance outcomes and potential remedies to reduce TC during ToD. The insights include unique contributions to team functioning under abrupt changes to meeting modality (i.e., from in-person to virtual and then to hybrid modality during the pandemic).

Characterized by fragmented nature and different conditions for every project, the Architecture, Engineering, and Construction (AEC) industry seeks remedies to achieve enhanced project outcomes including sustainability. Highly integrated project teams can overcome the obstacles by exchanging knowledge, goal alignment, and risk and reward sharing. Understanding the interaction patterns in project team networks can help implement desired practices to optimize sustainability outcomes. According to the influence theory, individuals are influenced by their interactions. They change their beliefs and practices in response to the people they connect with. Basic influence processes yield complex interactions and behavior patterns, and insights to these processes can help understand knowledge transfers in project team networks. The goal of this study is to examine the formation of knowledge transfer networks for sustainability in project teams via the lens of influence model and provide insights to complex organizational behaviors. To achieve study goal, we longitudinally collected e-mail exchange and archival data from an AEC project and performed Social Network Analysis (SNA) focusing on Energy and Atmosphere related interactions. The results showed that the exposure to the nodes with high-engagement levels and diverse expertise areas in a network positively impacts the sustainability knowledge transfers and engagement level of the individuals.

Interdisciplinarity is a critical component of creating a more sustainable built environment and improving student learning outcomes. Yet, learning interdisciplinarity and measuring it in higher education is challenging. This study implemented a diagnostic tool that allowed educators to measure how both disciplinary and interdisciplinary skills, knowledge, and values would grow in courses related to sustainable built environments. One school containing four different disciplines devoted to the built environment, was selected as the study area to collect emprical data. Pre- and post-semester surveys were conducted. Among 286 students, interdisciplinarity grew most among students within the school's majors, while among students from outside the school majors interdisciplinary learning decreased. Though disciplinary learning outpaces interdisciplinary learning the closer students are to earning their bachelor's or higher degrees. The result showed that students' knowledge, values and skills differ depending on the majors, their affiliational context, and the year of classification. The result suggests that the timing of teaching interdisciplinary contents should be earlier. The diagnostic tool which measured pre- and post-course disciplinary and interdisciplinary knowledge, skills, and values related to sustainable built environment would also be applicable in other courses in higher education.

The inherent characteristics of the Architectural, Engineering, and Construction (AEC) industry require experts from various organizations and disciplines to collaborate toward shared project goals. However, it is significantly difficult to connect individuals from diverse disciplines, backgrounds, organizations, and effectively integrate their knowledge. Social network approach addresses integration and know-how flow dynamics. However, there is a knowledge gap regarding how network topologies (i.e., parameters relating to nodes, ties, boundaries, and their placements across network) help improve knowledge transfers and optimize project outcomes. Thus, this study compares the network topologies of two similar AEC project teams. We collected e-mail exchange, observational, and archival data and conducted surveys from two projects during the construction phase and performed mixed methods of analysis. The study findings showed different network topologies between the two projects and one of the project networks evolved based on the project needs which yielded better performance. The results suggest that network parameters are not meaningful alone and engagement and network topology matter in measuring AEC project team performance. The results shed light on dynamic and multidimensional characteristics of knowledge transfer networks in complex project teams and contribute to the body of knowledge by providing a methodology in studying project teams in this light.

Architecture, engineering, and construction (AEC) project teams have a segmented organizational structure with subgroups, for example, designer, contractor, and owner. AEC projects are challenging to collaborate because they require those in different subgroups to address uniquely defined technical and functional contexts. The AEC industry often seeks integrated collaboration through organizational integration because the literature assumes that organizational structure determines collaboration structure. This study uses a network perspective to identify the inconsistency between the organization and collaboration networks through the data of email records from a $20 million AEC project with a typical fragmented organization. The analytical focus is on two network configurations: (1) a community structure through which subgroups are defined to attend to specific aspects of the project and then coordinated through ties between members of different teams; and (2) a core-periphery structure in which a relatively small number of members interact frequently in the core and then coordinate as each member of the core interacts with specific members of the periphery. Results provide evidence of integrated collaboration in fragmented project organization, indicating organizational integration is not a must to achieve integrative collaboration. The findings suggest implications to facilitate integrative collaboration: (1) efforts should focus on collaboration behaviors, (2) subgroups should adopt a dual-lead pattern, (3) subgroups should encourage non-high-profile members to function as cores, and (4) subgroups should ensure information sharing and prevent information overload.

Recently, architecture, engineering, and construction (AEC) project teams have begun to adopt the integrated project delivery (IPD) method with increased frequency. This adoption has been made on the assumption that IPD will lead to beneficial project team dynamics and outcomes. Although novel and potentially useful, we contend that the mere implementation of the IPD method does not in and of itself guarantee successful project team dynamics and outcomes. In specific, in this manuscript we argue that commitment disparities between project team members are problematic for numerous project team dynamics and outcomes: goal alignment, communication behaviors, and decision-quality. Using data from 21 IPD project teams, we show this to be the case. Results from this investigation suggest that when members’ commitment levels vary within project teams, goal misalignment, poor communication behaviors, and reduced decision quality are expected to follow. Moreover, results suggest that this is especially likely to occur when members’ commitment levels are low as opposed to high. This work contributes substantially to our understanding of project team dynamics in the AEC industry, especially as they relate to those delivered under contractually followed IPD. Although the IPD method holds the potential to be beneficial, these results suggest that such benefits accrue only if team members’ commitment levels are both high and uniform. In the absence of such conditions, problematic team-level dynamics and outcomes are expected to ensue.