Office Activity Taxonomy in the Digital Transition Era: Towards Situationally Aware Buildings

Abstract

In the context of the digital transition, office environments are increasingly shaped by flexibility, technological integration, and occupant-centered design. These transformations influence not only building operations but also the social dynamics and well-being of workers, thereby intersecting with the broader goals of socially sustainable design. To address this complexity, Building Management Systems (BMS) and Digital Twins must evolve from static automation to adaptive frameworks that recognize and respond to diverse workplace activities and social interactions. This study proposes a standardized taxonomy of office activities as a foundation for activity recognition and environment adaptation. A systematic literature review identified key activity categories and defining attributes, which were refined and validated through direct observations, diary logs, and semi-structured interviews in small, shared offices with open-plan workspaces. The resulting taxonomy comprises four main classes—Focused Work, Meetings, Shallow Work, and Resting—each defined by contextual attributes such as plannability, social interaction, number of participants, posture, modality, location, and duration. The framework supports the development of human-centric, situationally aware BMS capable of dynamically adjusting environmental conditions to promote comfort, well-being, and energy efficiency. By integrating user agendas and feedback, this approach contributes to more inclusive and socially sustainable work environments, aligning with the emerging paradigm of adaptive, human-oriented architecture.

1. Introduction

The COVID-19 pandemic produced new and multifaceted challenges in the management and design of physical workspaces [1,2]. In office-based environments, the widespread adoption of remote and hybrid work models has led to a fundamental shift in how employees approach on-site work [3]. This shift has significantly influenced expectations surrounding the physical work environment, with employees, now accustomed to the personalized comfort of home-office spaces, starting to seek more flexible, tailored, and comfortable spaces that better support their well-being and daily tasks [4].

As a result, organizations have reported rising levels of absenteeism and a growing hesitation among staff to return to conventional on-site office settings—especially when these spaces are perceived as overly standardized, disconnected from individual needs, and require long commuting times [5]. In parallel, recent reflections on Employment 5.0 emphasize that the future of work will increasingly rely on autonomous, creative workers who expect technologies and workplaces to adapt to their needs rather than the opposite [6]. Offices that fail to evolve in response to these changes risk not only reduced occupancy rates but also diminished employee engagement and productivity.

In this dynamic setting, a human-centric approach to workspace design has become increasingly important, moving toward the centrality of human perception as a critical component in the design process. Reciprocal interdependence exists between humans and the built environment, wherein individuals function as active agents within their surroundings while simultaneously being influenced by several multi-domain factors [7]. At the same time, treating the building as an isolated entity and focusing solely on its physical and technical properties is no longer sufficient. A human-centric approach enables us to reach more definitive conclusions on occupants’ environmental perception and well-being, integrating environmental data with physiological information [8,9] and subjective responses [10,11]. Such an approach prioritizes the needs, preferences, and well-being of occupants, promoting adaptable and adaptive environments that allow for the accommodation of different work styles and activities. Technological advancements, e.g., IoT-enabled systems, real-time sensing, and responsive environmental controls, now allow for the creation of dynamic spaces that adjust to changing conditions and user demands. These innovations support a shift from static, one-size-fits-all offices to intelligent environments capable of responding to the complex interplay between human behavior, personal and task-based requirements, and environmental conditions.

The function of an office derives from the activities carried out within it; therefore, it is essential to align the spatial and environmental characteristics of the workspace with the specific demands of these tasks [12]. An activity-based design approach enables the customization of office layouts, furniture, lighting, acoustic conditions, and thermal comfort parameters to suit the functional requirements of different tasks. For example, collaborative activities such as meetings or brainstorming sessions typically benefit from open, well-lit, and socially oriented settings [13,14], whereas tasks requiring deep focus or confidentiality may require more isolated, quiet, and controlled environments [15].

Moreover, as shown by Soriano et al. [15], the design of activity-appropriate spaces can have a positive impact not only on performance outcomes but also on the perceived well-being of occupants. A growing body of research has demonstrated that the relationship between environmental conditions (e.g., temperature, lighting, noise), individual preferences, and the nature of the activity being performed is a critical factor in supporting both comfort and productivity. Addressing this relationship requires a refined understanding of occupant needs, contextual variability, and behavioral dynamics.

In parallel, it is increasingly recognized that occupant behavior plays a crucial role in determining the energy performance of buildings. People’s interactions with systems, such as HVAC, lighting, or shading, can either enhance or hinder overall building efficiency. To address this, advanced systems such as Building Management Systems (BMS), Digital Twins (DTs), and predictive control algorithms are being employed to optimize indoor environmental conditions. These systems enable the integration of environmental sensor data with user-specific and contextual inputs, enabling real-time adjustments that consider not only ambient variables but also the intended use of the space and user feedback. For instance, Abdelrahman et al. [16] developed a digital twin with spatial–temporal proximity data to predict thermal comfort, obtaining not only an accurate incorporation of spatial context in the prediction, but around 20% higher accuracy than conventional models. Gnecco et al. [17], on the other hand, developed a digital twin to represent a test room system, integrating both the environmental data and the occupants’ information.

These factors highlight the need for a clearer definition and categorization of the diverse activities occurring within office environments, along with their corresponding environmental and design requirements. This involves identifying the optimal indoor conditions that best support each type of activity, thus enabling context-aware control strategies and more effective space utilization through developing an understanding of the activities’ main characteristics, priorities, and demands. For instance, the environmental requirements for holding a team meeting, such as enhanced ventilation, sufficient lighting, acoustic transparency, and more active layouts, differ significantly from those needed for focused individual work, which typically calls for lower noise levels, more stable temperatures, controlled lighting to minimize visual fatigue, and environments that are more stable and less subject to autonomous adaptation. Moreover, the different activities performed within an office could demand several other space characteristics, like the affordance of privacy or specific furniture. Collaborative activities, for example, could require a lower level of privacy than a virtual formal meeting with the company’s supervisors.

Furthermore, the reorganization of workplaces, particularly toward more adaptive, efficient, and health-supportive environments, is closely connected to the broader concept of sustainable development. Contemporary discussions show that sustainable development has evolved through historical, interdisciplinary, and systemic perspectives that emphasize long-term societal well-being and the capacity to adapt to complex global conditions. Works such as those by Manioudis and Meramveliotakis [18] and Klarin [19] illustrate this evolution, from the classical political economy tradition to modern sustainability frameworks. In this sense, improving workplace design and management aligns with sustainability’s core principles by promoting resource-efficient building operations, supporting human welfare, and responding to changing social needs. Reorganizing office environments to better match workers’ activities through situational awareness, adaptive systems, and evidence-based taxonomies contributes to sustainable development by fostering healthier, more resilient, and more productive work ecosystems.

Ultimately, the integration of human-centered design principles with adaptive, technology-enabled systems represents a promising strategy toward the creation of more supportive, efficient, and sustainable work environments. By acknowledging the diversity of activities and preferences within office spaces, organizations can better meet the changing expectations of employees while also achieving improvements in operational performance and energy efficiency.

To support energy efficiency and effective building system management, this study proposes a classification of activity types in office environments, along with a standardized vocabulary of their definitions and characteristics. This framework enables the incorporation of user agendas across diverse work settings, thereby improving the ability of building management systems to identify occupant activities and coordinate building operations accordingly.

2. Motivation and Purpose of the Study

Considering the rapidly evolving technological panorama, the categorization proposed here should (1) leverage shared vocabularies to learn and adapt to daily, weekly, and monthly schedules, proactively creating optimal environmental conditions tailored to occupants’ needs; (2) prospectively support designing a human-centered BMS (or similar system) capable of recognizing co-occurring activities and reacting accordingly to maximize the health and productivity of workers. The activity taxonomy proposed is therefore validated through real-world observational studies in office environments, ensuring accurate activity representativeness and characterization.

This work proposes a general categorization through a shared vocabulary of measurable working activities in office spaces, focusing on recurring activities while acknowledging non-recurring ones. Establishing common terms and definitions will help to delineate activity recognition through the job schedule and, consequently, enable situationally aware systems to adapt and create optimal environmental conditions tailored to occupants’ needs. Building on this general aim, the study also pursues the following specific objectives:

(a)

To identify and classify recurring working activities in office environments based on critical measurable parameters.

(b)

To define a shared vocabulary that standardizes terms for activity recognition across diverse office contexts.

(c)

To support and validate the proposed activity classification through real-world case studies.

3. Materials and Methods

The constantly evolving state of work environments since the industrial revolution demands a review of the concepts and requirements involved in maintaining healthy spaces and ensuring they align with new contexts and criteria, particularly in light of increasing flexibility and technology-driven work scenarios. This study introduces a taxonomy of recurring activities in office spaces, designed to assist smart office systems in taking adaptive actions to enhance occupant experience, with potential benefits for their health, satisfaction, and productivity. The taxonomy built here will provide a research-based framework to support consistent data collection and to inform the Situationally Aware Digital Twin. The most common classification of activities was identified through a systematic review of relevant literature, focusing on the definition of office activities and the attributes that characterize their occurrence. These findings were then compared with real-world events and reports from workers, obtained through observational studies, diary logs, and semi-structured interviews conducted in two distinct office settings.

The classification procedure followed a hierarchical structure to create definitions adaptable to diverse office contexts that are suitable for the office work activities described in the following paragraphs. First, the boundaries of the investigation were discussed by defining the types of space and work that are of interest in this study. Then, the activities and their attributes were characterized within the definitions proposed through the existing literature. The activity classification was then validated during an observational study in two different office spaces to demonstrate the applicability of the developed taxonomy in diversified contexts.

3.1. Definition of the Study Query

The research query was divided into four parts, each of which represented one dimension of the research question. The first part was space-related, limiting the studies found to the office environment. The second block guarantees that the research presents a people-centered view, putting occupants as the main characters and their activities as major players. The third section is associated with activity performance. Finally, the fourth part seeks to classify the activities carried out inside an office space. All these queries were combined into a single search, relying on the “TITLE-ABS-KEY” operator. The final terms and Boolean operators used are shown below in Figure 1.

The search was limited to peer-reviewed journal articles and conferences that were published in English. A temporal limit was not defined, considering that older studies could help to understand the transformation of the workspace and activities across the years. After the first search using the string presented in Figure 1, some additional works were added by the authors according to their previous experience in the field and their findings during the investigation for this study.

3.2. Data Selection and Activity Classification Criteria

This systematic search was performed on the Scopus database, and, after the removal of duplicate articles, it led to the identification of 136 studies. After the screening process, inspecting titles, keywords, and abstracts, 104 papers were removed from the database, being considered out of scope. The main reasons are listed below:

(a)

Studies not conducted in office environments

(b)

Studies that do not relate to any form of work activity performed in office settings

(c)

Studies focusing solely on individuals with pre-existing health conditions or diseases unrelated to their work activities or office environments

(d)

Studies addressing general workplace design strategies without linking these to specific, measurable office activities

(e)

Studies that only differentiate between sedentary and non-sedentary behavior without further classification or contextualization of work-related activities.

Some papers did not classify the activities performed in office spaces but provided important information about work dynamics and attribute assignment in different contexts. Thus, 16 additional works were added by the authors according to their previous experience in the field and findings during the investigation for this study. Figure 2 presents the framework for the articles’ selection.

Therefore, after following the procedure described, 48 articles composed the review database. A detailed analysis of the remaining papers explored the nature of the identified activities, their defining characteristics, and the environments in which they were carried out. Mokhtarian et al. [20] conducted a similar study focused on leisure activities. The classification of activities was performed with the assumption that understanding the nature of these activities can guide the coordination of building systems to enhance environmental conditions, spatial layout, and social interaction—ultimately supporting workers’ performance, comfort, and satisfaction. The chosen categories should be able to represent almost all work developed in the spaces described by the study boundaries, following the activity taxonomy indications described by Higgins and Safayeni [21]. At the same time, the classification should allow the identification of activities in the most precise way, without leaving residual uncertainty, but also not creating too many categories that could complicate the analysis and its association with modeling decisions. The review also considered the main stressors encountered while performing office activities.

The scope of the study was also clearly delineated by specifying the types of spaces included in the analysis. In this context, we first propose a general classification of the activities identified in the reviewed papers, organizing them into four main categories, which are presented in the Section 4. Following the initial categorization, all activities were further broken down into subcategories based on shared characteristics, with the goal of avoiding redundancy and enhancing clarity. A similar procedure was performed by Appel-Meulenbroek [22], categorizing the activities included in the analysis and their taxonomy. This process follows the approach suggested by Higgins and Safayeni [21]. Each subcategory was then assigned a final, representative label. The authors evaluated how activity-based office concepts are actually used after implementation, and how their design influences employee behavior, satisfaction, and productivity. This classification distinguishes environments according to their relationship with social, psychological, and job-related factors, considering whether activities are social or individual and planned or unplanned.

3.3. Real-World Observational Studies in Office Environments

To validate the activities defined through the bibliographical search, an observational study was conducted in two different office environments. Participants from office environments in Perugia, Italy, and Leuven, Belgium, completed a general questionnaire, disclosing demographic characteristics, working habits, and preferred workplaces. The spaces investigated and the procedure adopted are described in detail in the following two subsections.

3.3.1. Characterization of Case Studies

The first environment consisted of five small, shared offices, each one accommodating from one to three people and featuring similar dimensions and sun-exposed façades. The structures were located on the first floor of a two-story building at the Engineering campus of the University of Perugia (Italy) (Figure 3) [23], belonging to a humid subtropical climate (Cfa) [24]. The main environmental parameters in the offices, such as air temperature, illuminance, CO₂, VOC, PM concentrations, and air movement, are continuously monitored. Occupants are able to adjust mechanical ventilation settings in the room to suit their preferences, in coordination with the university’s district heating and cooling system. Local regulations dictate the daily operating schedule for the HVAC system during summer and winter periods, allowing thermostats to be adjusted within a range of ±3 °C from the global setpoint established by the centralized system. The windows and the lighting systems in the offices are operable by the users, who also have the support of Personalized Environmental Comfort Systems (PECS), to enhance their comfort and well-being locally.

The second office environment included two open-plan rooms on the ground floor of a multi-story building, part of the KU Leuven campus (Figure 4): A and B, which were 110 m² and 210 m² in area, respectively. The construction was located in Leuven (Belgium), part of a Temperate Maritime climate (Cfb). These two office spaces are shared by the same group of workers, who can freely choose which room and desk to work at each day. Office A includes 18 desks, grouped in 5 islands, while Office B comprises 21 desks, grouped in 4 islands. Both offices in KU Leuven have windows located next to the island desks. These spaces are suitable for research and development work, considering their typical open-plan office setting. The two offices at KU Leuven, equipped with a total of 39 desks, provide a workspace for two-thirds of one of the department’s research teams. Environmental parameters in the offices were not measured during the current study. Participants can change the lighting in both offices and the shades only in office A.

Data collection was conducted over a period of 9 consecutive working days by 2 researchers and aimed to identify workers’ routines, experiences, and interactions during a full cycle of work, also acknowledging repeating and non-repeating activities, which are better detailed in the following subsection.

3.3.2. Field Study Data Collection

The procedures adopted in Perugia and Leuven varied slightly, reflecting the differing availability of the respective research teams for data collection. Nevertheless, the core objectives and methodology were preserved to ensure comparability between the two office structures and alignment with the study aims outlined in the previous chapters. The methods used at both sites were semi-structured interviews, a periodic questionnaire, and diary logs, that can be accessed in Table S1 of Supplementary Materials. In Leuven active participant observation was also carried out, which will be described in the next sections.

• Participant observation

This method was applied only in the KU Leuven offices, described as a prolonged observation and participation in occupants’ routines and everyday practices. Participants’ daily work activities and interactions within the shared workspace were observed by two researchers, who were present in the same office for the first two days and then were present in separate offices. During the observation, researchers were taking notes, which, besides general observations of participants, included the researchers’ personal experience in the office as well.

• Semi-structured interviews

The interviews were conducted privately in Perugia and Leuven, involving one researcher and one participant, aiming to gather insights into the workers’ preferences, office experiences, job tasks, primary activities, daily routines, and the influence of external stimuli on their work experience. Semi-structured interviews were conducted, following an overall structure with three key sections: introduction, social impacts, and physical impacts. These sections were further divided into specific topics. For example, in the social impacts section, topics included relationships with coworkers, preferred places of work, organizational culture, privacy, and others. In the physical impacts section, topics included workplace setup, noise, lighting, etc. The interview structure can be checked in Section S1 of Supplementary Materials. Each interview concluded with a short final reflection. A total of 22 semi-structured interviews were conducted with office occupants—6 in Perugia and 16 in Leuven. Eight interviews took place in person at the Leuven site during the fieldwork period, while the remaining interviews were held remotely in the following weeks using online meeting platforms.

• Questionnaire

In both work sites, an online general questionnaire was distributed and remained open for responses over a two-week period—in Leuven during the weeks of the observational study and in Perugia after the first interview section. The questionnaire consisted of 41 questions divided into several sections, including demographics, workplace characterization (offices A or B, other areas), and a comparison between offices A and B. Participants responded to the questionnaire at different times during the two weeks. The questionnaire served as a reference point for the semi-structured interviews.

• Diary logs

The diary logs were flexible templates that were filled in by the occupants, documenting their daily activities alongside contextual details such as time, location, people involved, and a rating of how well the environment fit each individual activity. The form also included a section for comments, enabling participants to share a one-day diary of additional observations or reflections. This format gives the subject the opportunity to more freely express their understanding of the environment and their priorities. This procedure was repeated twice in Perugia’s offices, with approximately one and a half months between each session. In Leuven, 9 participants were required to fill out the diary log templates throughout one of their workdays and to capture any observations, changes, feelings, or thoughts as they occurred. Days were arbitrarily chosen by the researchers based on the general atmosphere in the office.

• Informed consent

Participants in the experimental campaign were informed about the study objectives, the parties involved, and the potential risks and benefits of taking part in the different phases of observation, including diary logs, participant observation, and interviews. No physical measurements were taken at any point during the study; only subjective and demographic information provided by the participants was collected. They were also made aware that they could withdraw from the study at any time without any consequences. All this information was detailed in a written informed consent form, which participants read and signed before beginning their involvement, authorizing the anonymized and non-traceable publication of the data they provided.

After the data collection process, the activities gathered from the literature review and from the field, namely diary logs, interviews, and observations, were systematically clustered according to the predefined classes of activities, such as Focused Work, Meetings, Relaxing, and Shallow Work. This process involved a comparative examination of the different data sources to identify recurring behavioral patterns and cross-validate the findings. Subclasses were then developed by grouping activities with similar functions and environmental requirements, assigning a representative label to each. To support the reliability of this qualitative coding, key distinguishing attributes, defined through the combined analysis of the collected database and the reviewed literature, were consistently assessed for each identified activity. These attributes served as a guiding framework for refining subclasses and ensuring that the resulting taxonomy was coherent with empirical observations and aligned with established theoretical approaches.

4. Results

This section describes the outcomes of the research. First, the dataset from the literature review is described, together with the definitions adopted for office spaces and office activities. Insights from the literature are used to establish conceptual boundaries and to support the initial classification of activities into classes and subclasses. The empirical study, comprising observations, diary logs, and interviews across both office settings, serves to validate and refine these boundaries and to expand the list of reported subclasses. Building upon both sources, the authors then develop a theoretical operational framework for activity recognition and BMS decision-making, including the baseline attributes assigned to each subclass and the adaptive workflow.

4.1. Describing the Dataset

This section analyses the 48 papers, of which 27% are conference articles, selected through the search structured following the rules presented in Section 3.1. All studies investigate activities conducted in office environments, although the specific type of office space is not always clearly defined by the authors, and its definition depends on the study’s aim. In several cases, the primary focus is on the “office worker” rather than the characteristics of the environment in which they operate. From the selected studies, seven were performed in open-plan offices, six in single offices, six in shared offices, one specifically in a home office, sixteen performed comparisons between or considered two or more office types, and seventeen mentioned “activity-based workplaces” or related terms. The latter are workplaces that allow employees to choose where and how to work, based on their tasks and personal preferences [25]. Moreover, 14 studies did not specify the office type.

Considering the countries of origin, the most represented were the Netherlands with 10 publications, followed by the United States with 7, and both Germany and Sweden with 4 each. Less-represented countries included Bangladesh and Italy with 3 publications each, and Australia, Finland, and Ireland with 2 each. Over the last 10 years (2015–2025), 34 articles from the sample were published, compared to 14 published between 1990 and 2014. This trend reflects a growing interest in the field in recent years, likely influenced by evolving workplace dynamics and emerging approaches.

Subsequently, keyword occurrence was investigated, based on both the authors’ keyword lists and the abstracts of the selected papers. A minimum occurrence threshold of 3 was delimited, resulting in 125 terms. After applying a relevance score cutoff of 60%, 74 terms were retained. The most frequently cited keywords include “office” (12 occurrences), “workplace” (7), “work environment” (6), and “office environment” (4). Related terms such as “workplace design” and “office concept” each appear four times, while “office design” is cited three times. In terms of activity monitoring, “activity classification” and “office activity recognition” are cited five and three times, respectively. Keywords reflecting characteristics of office activities include “performance” (11), “preference” (7), “satisfaction” (6), “collaboration” (5), “comfort” (4), “attention” (3), “privacy” (3), and “concentration” (3). Additionally, several terms related to occupant health are present, such as “risk” (five), “health” (four), and “health care” (three). The terms that occur in more recent publications are frequently connected to data analysis and model evaluation, such as “dataset”, “precision”, and “accuracy”.

4.2. Conceptual Boundaries of the Investigation

This section examines various concepts and topics in the body of the defined dataset that are relevant to the work activity classification developed in this study. These conceptual boundaries include health risk factors, environmental influences, and key elements affecting activity execution and performance. Additionally, the types of spaces and their fundamental structures are outlined, along with the activities typically conducted within these settings, as informed by the literature and the observational studies conducted within the research framework.

To provide clarity and context for the investigation, specific definitions are introduced in the next sections.

4.2.1. Office Spaces

The simple definition of office has proven to be difficult in the past for other authors, as Dodswell explains [26]:

“Viewing the office as a place where white collar work is conducted or as a set of functions and activities whose output is written and oral communication is likely to lead to an unacceptably narrow focus”

In recent years, the concept of the office and office work has changed a lot, considering all the innovative solutions, design proposals, and technologies created to facilitate processes, optimize time, and increase the productivity and satisfaction of workers. The changes driven by globalization and the expansion of digital technologies have had profound impacts on the nature and practices of work organizations, corresponding with the increase in flexibility and mobility [27]. In any case, the social context of an office is still complex and should be clearly delimited to clarify the purpose of this study [28].

This study narrows the general concept to “Office spaces”, referring to individual or shared environments with unique or multiple uses where office activities can be performed. Nevertheless, an office space should also guarantee some basic structures or supports to enable the user to use their own tools to perform digital knowledge work, plus offer the possibility of environmental adaptation or the inclusion of smart systems. It mostly concerns conventional offices, as the validation process was performed in one small, shared office and in one open-plan office.

Considering these boundaries, the full complexity of the workspace’s role in task performance is often overlooked by managers [29]. For instance, hot-desking models can offer workers greater flexibility in terms of location and scheduling while also encouraging social interaction. At the same time, such setups may undermine employees’ sense of belonging within the workspace [30]. Emerging models of work aim to align the specific demands of diverse job activities with appropriate environmental settings [22]. A thoughtfully designed office that supports collaboration or creativity can play a vital role in task success [13] and may be a key factor in attracting employees back to the workplace in the post-pandemic context [31].

Offices can also be defined in terms of physical organization. Considering the different types of office spaces, we acknowledge firstly the widely recognized definition by Neufert [32], which divides offices into cellular offices, group offices, and open-plan offices. Cellular offices comprise one to six people and mostly support tasks requiring concentration and minimal interaction (low interactivity and high complexity); group offices can house 6 to 20 work stations, being more adapted to collaborative teams, handling complex tasks that require information exchange; and open-plan offices can cover more than 20 work stations to accommodate large teams, suiting routine or less cognitively demanding tasks, whether interactive or not [15]. However, in light of emerging work trends, two additional workspace types should be considered here: coworking spaces and home offices. Coworking spaces are part of the broader post-digital transformation in society and are typically described as digitally connected environments where freelancers and other flexible knowledge workers can access shared workstations, called “hot desks”, and office amenities on a temporary basis [33]. Home office refers to a flexible work model that allows employees to perform their job duties remotely, typically from their own homes, using information and communication technologies. It emphasizes autonomy, performance, quality of life, and social interaction, while requiring clear frameworks for management, legal compliance, and evaluation [34]. In this matter, we consider also the concept introduced by Ray Oldenburg [35,36], known as the “third place”, comprising workspaces that are alternatives to home (“first place”) and conventional offices (“second place”), e.g., coworking spaces, cafes, and other “hybrid spaces”.

Considering employees’ perspectives, Appel-Meulenbroek et al. [31] carried out a study in which office workers were asked to indicate their preferences based on specific attributes of the work environment and other influencing factors. The aim was to identify optimal conditions and preferred workplaces across three scenarios: communication, balance (50/50), and concentration. The authors found that, in hybrid workspaces, employees tend to fall into two groups: those who prefer working in the office—typically full-time male employees with short commutes and communication-focused roles—and those who favor working from home—generally part-time female employees in administrative roles with longer commutes. An additional insight from the study was that, although noise and distractions are major concerns in office settings, workers do not typically address these issues through complete isolation, avoiding even the use of designated “concentration rooms”. A balance between focused work time and the possibility of communication should be established to guarantee workers’ satisfaction and improve their performance.

The environmental and layout characteristics in office spaces could also influence workers’ satisfaction and productivity, as pointed out by many researchers [37,38]. Bergefurt et al. [39] conducted a study on how the physical workplace affects employees’ mental health, analyzing 133 articles. They found that most of the reviewed studies focused primarily on office lighting characteristics, often neglecting factors like biophilia and views, and largely relying on subjective data. Other physical characteristics that may influence occupants’ choices include the accessibility and suitability of workspaces, such as their overall crowdedness, the presence of dedicated concentration and meeting areas, and the provision of adequate equipment to support job tasks [13]. The adoption of ergonomic office workstations, for example, may influence workers’ sitting behavior and, consequently, their health and well-being [40]. Building managers in knowledge-intensive organizations should pay more attention to adapting the environment to the activities performed, in order not to harm workers’ health and the organization’s performance goals [41].

After defining the space type investigated, the activities of interest should be established, considering the close relationship between a space and the actions performed within it. Raymond and Cunliffe stated in 1997 [42] that workplace decision-making fundamentally revolves around activities. As already described, space can give a sense of meaning to an activity or simply support its execution, allowing interactions or intense, focused work activities [43]. In the next section, the type of activities investigated here will be characterized.

4.2.2. Office Activities

In this study, we are investigating “Office Activities”, which refers to sedentary, intellectual activities, performed predominantly using computers and digital tools in an office environment. This type of work often requires sustained cognitive focus and prolonged screen time, with minimal physical activity and high reliance on technology to progress with the work. In addition to core work tasks, “shallow work” [44] should also be considered, comprising tasks that, while less cognitively demanding, are essential for task management, facilitating communication, and coordinating schedules, performed primarily in the office environment. These activities, though often overlooked, contribute to a more organized workflow and can ultimately lead to greater freedom, productivity, and effective time use. Finally, the activities explored in this study are primarily associated with knowledge work, a concept introduced by Peter Drucker in the 1950s [45], referring to the creation of value through the development and application of knowledge, information, and understanding [46], rather than through manual or physical labor. While the authors recognize that all forms of work—including those involving physical tasks, such as in construction sites—require a combination of experience, theoretical understanding, and practical skills, this research focuses exclusively on activities typically carried out in predominantly sedentary, office-based environments.

Additionally, we focus on those activities that occur multiple times throughout the week and form a regular part of the workers’ routines. This regularity is particularly important for two main reasons: first, because building management systems (BMS) or situationally aware systems may be able to detect these patterns; and second, because it enables participants to develop familiarity with the activity, allowing them to reflect on their feelings and sensations during its execution, evaluate how the surrounding environment affects their performance, and make adjustments either to the activity or to the environment, if needed. The “office experience” [47] significantly influences how activities are carried out, shaping the functional, social, and symbolic dimensions of interactions and dynamics among employees.

Another dimension of office activities is the task, which, according to Byström and Hansen [48,49], is “a workable analytical unit of human activity, which brings the level of explication close enough to cater for individual actions and their consequences”, i.e., activities can be broken down into small components, called tasks. Essentially, activities operate at a macro level, encompassing multiple tasks, which function at a micro level. An activity represents a broader goal-oriented process, while tasks are the specific, actionable steps that contribute to completing that activity. For example, in research work, the activity of writing a paper may include several tasks such as compiling the dataset, reviewing literature, analyzing data, structuring the presentation of results, and editing paragraphs (Figure 5). This distinction allows for a more granular analysis of work processes and individual contributions within a broader framework.

4.2.3. Observational Study Considerations: Diary Logs, Semi-Structured Interviews, and Active Participant Observation

In the observational study conducted in the two different types of offices, which supported the validation of the activity classification and characterization in this study, a combination of qualitative and quantitative methods was applied, as described in the methods section. A subjective analysis of the diary logs, semi-structured interviews, and participant observation carried out in Perugia and Leuven is reported in the following section.

In the diary logs, all participants, both in Perugia and Leuven, mentioned focused work or a correlated terminology, which was also the longest activity in terms of duration. Focused work was mostly understood by participants as individual computer-based activities, preferably without any distractions, considering that most of the workers were involved in research activities. The second most frequently mentioned activities were having breaks and attending meetings. When it came to breaks, participants mentioned specifically coffee breaks, which mostly took place outside of the office, in other areas of the building, and were usually collective rather than individual. Lunch breaks were predominantly considered to be collective activities.

Since the researchers observed the participants filling out the diary logs in Leuven, they can verify that even people who did not specifically mention lunch breaks in their diary logs actually did have lunch with colleagues. There was probably some misinterpretation in the Logs compiled by participants, as they registered mainly activities that had a close relationship with work. Other corresponding activities to having breaks are filling bottle, having a break in general, toilet break, and tea and fruit break. Most of those activities happened outside of the office and are therefore not considered in the activity taxonomy in this study.

Attending meetings was also a very common work activity among participants. All the participants attended different meetings, and all of them were outside of the office, in Leuven’s case, but still in the same building. Most of the meetings happened in the meeting room next to Office A. The Perugia study’s occupants usually took meetings, online or in-person, in their own small, shared offices, or in a colleague’s office with a similar configuration. Less frequently mentioned activities were setting up desk and cleaning up desk, which can be categorized as shallow work.

In the semi-structured interviews, the frequency of mentions showed slightly different perspectives. Focused work only emerged as the third most frequently discussed topic. This activity was something that was done throughout the day, both solo and with coworkers. Workers also viewed their productivity as being higher when they worked for most of the day on a specific task. Having a space where practitioners can complete focused work with minimal interruption was critical to them. In Perugia, some participants complained about the noise from other office mates and from outside, in addition to many interruptions coming from colleagues from other offices. Leuven was a shared, open-plan office space, and participants considered fewer people to be better when describing the number of people surrounding them. In both offices’ configurations, the importance of a quiet environment to perform focused work successfully was emphasized.

Relaxing was the most frequently mentioned topic, including all kinds of breaks, which does not mean that workers spend a large portion of their day on a break, but rather that those breaks have value for participants, and this value is mainly social. Participants in Leuven also mentioned the need for a higher-quality area in which to take breaks. There was also considerable discussion about the lunch break, considered the main daily break/social part of the day for every employee. Employees usually eat lunch in a specific place, designed as a place to have meals, in both Leuven and Perugia. It was found that big groups often gather there in both groups investigated, sometimes even from eight to fifteen people, when the space is busiest. Everyone does not eat at the same time, because participants have different schedules. They join a group that is already sitting when they can. Employees can sit and talk quietly, which creates informal bonds. People discuss both work topics and personal matters. Some people skip lunch, especially if they do not want to talk or have too much work, which does not happen often. Lunches are thus key to the social dynamics in the work environment.

The Coffee break theme also led to interesting conclusions: some of the participants do not even consume coffee, yet they take coffee breaks because these are usually geared towards socialization, networking, and the feeling of belonging. Coffee breaks are therefore central to their social interaction at work. Most of the staff, apart from a few, consume their coffee during the first hour after coming into work and subsequently in the afternoon after lunch. Arguably, both coffee and lunch breaks give their day a structure, because rituals related to coffee and lunch take place approximately at the same time each day. Coffee breaks, being a counterbalance to socializing, establish tension between concentration and social contact. Some employees are often in the middle of deep work when their peers call them to have a coffee, and they must choose whether to break their flow or not. Having already missed the coffee break with peers, they will feel behind, shattering their sense of belonging.

The second most mentioned activity was attending meetings, which were divided into five different subcategories based on participants’ quotes, for better clarity: informal meetings in person, formal meetings in person, formal meetings online, online communication, and meetings in general. The general meetings subcategory contained cases where participants talked about meetings with internal or external co-workers; they did not state the kind of meeting, but referred to them generally or mentioned them casually. Formal meetings occur in person in the meeting room next to Office A, in the case of Leuven (open-space office). This room is often busy, so people meet elsewhere in the building. They use classrooms and labs, if they are available. The meeting room is the place people prefer for formal meetings in person. The informal meetings in person subcategory included not only pre-arranged informal meetings but also spontaneous social interactions of varying duration. These spontaneous encounters fall into this subcategory because conversations often shift, sometimes only briefly, to work-related topics. However, most of these interactions revolve around non-work-related subjects, which employees especially value and appreciate as a mental and cognitive break from the workflow. The formal meetings online are occasionally held from participants’ offices in Leuven when their participation in the meeting is passive, i.e., listening. When actively required to contribute, they prefer to go to the meeting room or another quiet area in the building. The final subcategory is online communication, which refers mainly to interactions through various digital applications that allow employees to quickly discuss matters that would otherwise require them to set up a meeting, or to use the apps themselves to arrange one. In Perugia, as workers do not have a specific room, they usually take all meetings in their own office space.

The least frequently discussed category was learning activities, with references to teaching and research. Even when encouraged, participants did not pay too much attention to the topics. At the end, for the sake of definition and simplicity, this category was incorporated into Meetings, when a collective activity was performed to share information, and into focused work, when learning involved individual activities, such as a reading activity. The second least frequent activity was shallow work, which included the settling-in phase after arriving in the office, and phone calls, although it is not entirely clear whether all the phone calls were work-related. Some participants mentioned performing shallow work between other bigger tasks, as they usually do not require much concentration.

In the context of active participant observation, performed only in the Leuven offices, the meetings category also included social interaction that occurred mainly in the office or began there and continued outside the office, for instance, when a conversation was initiated in the office and then moved elsewhere so as not to disturb colleagues. These were mostly brief exchanges between coworkers, typically taking place at their desks. The participants would rise to approach colleagues they wished to speak to, and a few would even roll over to them in their chairs. There were other days—particularly when the staff was struggling to achieve deadlines for reports and similar tasks—when the amount of social interaction in the office decreased significantly.

Focused work was identifiable during participant observation primarily as computer-based activity and in interviews through participants’ own accounts of engaging in it. As a result, there were no detailed observations or systematic records specifically related to focused work. However, participants still spent most of their time on it, as evidenced by data collected through other methods. Even though the participant observation method did not capture much in terms of focused work, it was nevertheless the most frequently performed activity in the office during the fieldwork period in Leuven.

Within the relaxation category, the most common work activity that was monitored was the coffee break, followed by the lunch break. These were two of the few activities where researchers engaged with the participants. As previously stated, coffee breaks and lunch breaks are the most crucial points for the reinforcement of social solidarity among the collective. For both, there was always someone in the office who would initiate, and the others would follow. In extending the invitation to others to join a coffee break, it was not even unusual for the initiator of the break to hold their coffee cup high in the air so that the people wearing noise-canceling headphones—who otherwise might not hear the invitation—would be able to still see the invitation. This movement served as a tangible cue for the coffee invite.

For lunch, workers depended more on the hour, and no special effort was made to summon others. A loose rule was that lunch would be at 12 P.M. in Leuven, and those who could not make it precisely at noon would just arrive later when they could. In Perugia, no specific time was defined—workers usually gathered in the range from 1 P.M. to 3 P.M. Beyond the office coffee breaks and lunch hour, two other behaviors were seen: eating in the office and work breaks. Eating in the office was usually performed at the office desk, and consisted of little snacks such as nuts, chocolate bars, or fruit bits. Whenever subjects took a short break while at work, they always exited the office.

In addition to the aforementioned activities, two more were observed to fall under shallow work: preparing the desk for work, which included cleaning and disinfecting the desk before use, and using the phone. However, the specific purposes for which the phone was used could not be determined through participant observation due to privacy concerns.

In sum, according to the observational studies performed in both Leuven and Perugia, the most frequently observed work activities during participant observation fell into the categories of shallow work, meetings, relaxing, and focused work. The determination of activities’ subclasses and their attributes is described in the next section.

5. Determination of Office Activity Taxonomy: Classes and Subclasses

This categorization is informed by observations of office work dynamics and an analysis of employees’ experiences and activities during the observational study conducted in two real office environments, as detailed in the previous section, and is in alignment with the reviewed literature. Duffy and Powell [50], for example, introduced an activity division based on the nature of workspaces—namely, group processes, individual processes, transactional knowledge, and concentrated study. This framework was later refined by Hascher et al. [51] based on the activities performed. Inspired by these models, our proposed classification aims to encompass the widest possible range of office-based activities. Activities were grouped into five overarching classes, representing the most common types of actions performed within office spaces: Focused Work, Shallow Work, Meetings, and Resting periods. Activities that took place outside the office, such as off-site lunches, decompressing walks, or outdoor meetings, were excluded, even if they occurred during working hours. The definition of each activity class is described in

Table 1. Activity classes and definitions.

General Classification	Definition
Focused Work	Period dedicated to activities that require concentration on a specific task, ideally free from interruptions or distractions. Often defined by single-minded attention to complete complex, detail-oriented, or creative tasks
Meetings	Online or in-person interactions involving two or more individuals, typically within a professional or organizational context, aiming to exchange information, discuss ideas, make decisions, or foster collaboration
Shallow work	Non-cognitively demanding, logistical-style tasks, often performed while distracted. These efforts tend not to create much new value in the world and are easy to replicate.
Relaxing	Informal and low-pressure breaks that give employees a mental and physical interruption from work tasks. These activities are important to maintain well-being, reduce stress, promote team-bonding, and enhance overall productivity.

.

The main attributes that supported the differentiation between the activity subcategories were as follows: social interaction, posture, possible health risk, duration, complexity, and privacy. The specifics for these attributes will be investigated based on the articles gathered in the current Section 5.

The structured classification of these activities was also developed to consider the recognition of occupant activities by Building Management Systems (BMS) and support the adaptive control of building systems based on specific user needs. Each general class comprises multiple subclasses that group activities with shared characteristics together, as identified in both the literature and empirical observations. These activities exhibit similar environmental requirements and behavioral patterns, making the classification useful for context-aware building system responses. Activities that, considering the description given by participants or the articles reviewed, could be classified into more than one subclass were duplicated. Figure 6 presents the activity classes and subclasses, while

Table 2. Activity classes and subclasses, assigned through the activities gathered through the literature review and observational study.

General Class	Subclass	Activities Gathered in the Literature Reviewed	Activities Gathered in the Observational Study
Focused Work	Coding	Write code; write script; write tool	Coding; simulation/technical activities; data analysis; catch-up results annotation; work desk time; data processing and visualization
	Writing	Writing; Write an essay; generate text; sitting/standing and typing	Writing master’s thesis; working on literature review; writing a journal paper; report writing; research writing and documentation; research writing; research documentation
	Reading	Reading; sitting/standing and reading	Reading review papers
	Focused Individual Tasks	Private work; individual work; individual low/high concentration work; retreating alone; individual work tasks requiring concentration	Testing sensors; practicing a presentation; focused time
	Visual–Cognitive Tasks	Creative work	Preparing a presentation; graphic edits; making slides (presentation)
	Cognitive–Administrative Work	Coordinative activities	Engaging in administrative tasks; contract reading
	Research-Oriented Tasks	Gathering info	Formalizing/working on research outline proposal; working on a PhD
	Review-Oriented Tasks	-	Correcting tables of a review paper; correcting journal paper proof
Shallow work	Routine Office Tasks	Behind the computer; routine at the computer; general desk work	Office work behind desk/laptop/screen; checking calendar; making poll; sending emails; answering surveys
	Physically Supported Office Work	Archiving; sitting/standing and sorting paper; print; staple	-
	Digital-Concentration Office Tasks	Document management	Correcting minutes
	Setup and Closing Tasks	-	Cleaning up desk; packing to leave; locking my bag in my drawer; setting up desk, laptop, and peripherals; wrapping up work
Meetings—In-person/Online/Virtual meeting	Collaborative and Coordinative Meetings	Formal meeting; coordinative activities; ad hoc collaborative activities; mandatory collaborative activities; collaborating	Catch-up with supervisor; meeting with experts; checking purchases with a co-worker; discussing purchases/contracts
	Informal Collaborative Activities	Informal meeting; group work; answering questions; creative work; called collaborative activities; teamwork; help finding information/giving info; informal meeting; conversation at the meeting table	Brief consultation with a colleague
	Strategic Collaborative Work	Deep collaboration;	Project alignment; meeting with a colleague about joint collaboration/project
	Spontaneous Collaborative Interactions	Spontaneous communication/meeting; on-demand meeting; change meeting; unplanned meeting; job networking	-
	Presentation Activities	Presenting; presentation	Presenting my work
	Training and Learning Sessions	Workshop meeting; learning activities	Book club; conference meeting; online journal club
	Brief Telecommunication Work Interactions	Phone conversation; telephone use; phone calls; video calls
Resting	Social Interactions and Breaks	Internal conversations; conversations at the entrance; conversations between two far apart offices; informal talk; desk work interaction; spontaneous communication; self-reported leisure time; Intermediate activities that interrupt work (lunch, break, drink, toilet, printer, mailbox, smoke, sport, etc.); lunch; coffee	Social break; informal chatting
	Brief Telecommunication Breaks	Phone conversation; telephone use; phone calls; video calls	-
	Occupational Physical Activity	Sport/exercise; occupational physical activity	-
	Individual Break	Sitting/standing quietly	-

shows the information gathered from the literature and the observational study that originated each subclass.

While cultural and organizational norms may shape how frequently certain activities occur or how they are socially interpreted, the fundamental nature of office work tends to remain consistent across countries, as observed in the field studies and in the reviewed papers. Activities such as individual, focused work, collaboration, meetings, digital tasks, and breaks are common to most contemporary workplaces due to globalized work practices and shared technological infrastructures. For this reason, the taxonomy is grounded in core, observable activity types and measurable attributes that show strong cross-cultural stability. Nevertheless, the framework is designed to be adaptable, allowing researchers or practitioners to account for local cultural nuances when applying the taxonomy in specific geographic or organizational contexts.

To maintain usability while ensuring operational relevance, the taxonomy is structured around four broad activity classes that facilitate an initial identification by building managers and automated systems. However, the inclusion of multiple subclasses is essential to differentiate the distinct patterns of spatial, environmental, and behavioral requirements that directly affect how adaptive architectural systems respond. To support practical implementation, each subclass is linked to a set of characteristic attributes, which may be input into the Building Management System to help identify the most probable activity and suggest suitable subclasses to the user when selecting a space. This structure balances simplicity at the class level with the level of detail required for accurate and meaningful adaptation at the building-operation level.

5.1. Definition of Office Activity Attributes

As already mentioned in the methods section, in order to be able to identify different activities and their needs in an office space, enabling support for the completion of these activities and a better environment for workers, some attributes should be defined. Mokhtarian et al. [20] performed similar work for leisure activities, defining their taxonomy into five general dimensions: location, time, social context, traits, intrinsic to the activity, and the benefit/cost trade-off. In that case, the classification supported the authors in understanding the role of information and communication technologies (ICT) in individuals’ leisure choices and associated travel behavior.

Identifying workplace activities and their key attributes is essential for designing well-planned and personalized office environments that align with occupants’ needs and task requirements. Several studies have explored strategies for the automatic recognition of office-based activities. For example, Cha et al. [52] employed an accelerometer to detect seven predefined static and dynamic office activities performed by participants in a controlled setting for five-minute intervals. By training an ensemble classifier model, the authors achieved an accuracy of 96.1%. Other researchers have applied similar techniques to assess sedentary behavior in office contexts, recognizing it as a potential health risk for workers [53,54].

Based on the outcomes of both the reviewed literature and observational studies, a set of relevant attributes was identified in relation to the primary requirements of Building Management Systems (BMS) and recognition frameworks. The system under consideration is expected to process scheduled occupancy data for a given space and infer the specific environmental requirements necessary to support the intended user activities:

• Plannability: refers to the degree to which an activity can be anticipated and scheduled in advance. Activities may be broadly categorized as schedulable, i.e., those planned with prior notice, or spontaneous, which occur without pre-established timing. This attribute plays a crucial role in determining the priority level of activities for the coordination of BMS. Schedulable activities allow for proactive adjustments in environmental settings (e.g., lighting, moveable panels, shading systems), while spontaneous activities, due to their irregular and unpredictable nature, present challenges for pre-emptive system responses. However, when supported by intelligent automation, spontaneous activities can be detected in real-time, to then be validated through occupant feedback with short surveys or prompts, enabling adaptive calibration and data enrichment for future occurrences.
• Social interaction: activities can be classified as individual or collective. For collective tasks, the number of participants that are physically involved (not online, in the case of meetings, for example) should be specified to ensure appropriate spatial configuration, including adequate furniture and physical space. Individual activities may require higher levels of concentration and, consequently, lower tolerance for disturbances or background noise.
• Modality: refers to the mode of participation in an activity and is particularly relevant for collective tasks. It determines whether the activity is conducted physically, online, or in a hybrid (mixed) format, where some participants are present in person while others join remotely. This classification informs the necessary spatial and technological configurations required to support the activity. For instance, online activities typically demand minimal physical preparation—limited to accommodating the individual respondent with appropriate digital tools (e.g., computer, internet access). In contrast, physical and hybrid modalities necessitate the arrangement of suitable environments, including adequate seating, equipment, and acoustics, to ensure effective interaction among participants. Understanding the modality of activities allows Building Management Systems (BMS) to tailor environmental responses in alignment with the actual spatial demands of occupants.
• Posture: This refers to the body position adopted during the activity and can be categorized as seated, standing, or flexible. Seated activities typically demand ergonomic furniture and workspace accommodations, whereas standing or flexible positions may place fewer requirements on the physical environment. Activities classified as flexible are generally adaptable to the available space and do not impose specific spatial or furnishing demands. Offices equipped with movable desks should be considered “flexible” for individual activities.
• Duration: refers to the expected time frame of the activity, including its start and end times. This attribute is essential for scheduling building systems (e.g., lighting, ventilation, thermal control) in alignment with occupancy patterns. When possible, durations should be specified to support predictive system responses and efficient resource allocation.
• Complexity: indicates the cognitive or operational demand required to perform the activity. It can be classified as low, moderate, high, low to moderate, moderate to high, or not specified. Higher complexity levels may imply the need for enhanced environmental support (e.g., optimal lighting, minimal distractions, or ergonomic conditions) and lower tolerance for disturbances or background noise.
• Privacy: indicates the degree of confidentiality or seclusion required for the activity. It is similarly classified as low, moderate, high, low to moderate, moderate to high, or not specified. Activities with higher privacy needs may require acoustic insulation, restricted visibility, or spatial separation from other occupants.

Previous research has sought to differentiate the attributes of work activities to improve both workplace dynamics and office design. These studies offer important insights into the characteristics that should guide the development of responsive BMSs. One such characteristic is the degree of collaboration involved in an activity. While distinguishing between individual and collaborative work is useful, treating this classification as a strict binary can oversimplify the complexity of workplace behaviors. Instead, a more nuanced understanding is needed—one that explores the spectrum of collaborative strategies, contextual challenges, and degrees of interdependence that exist within and between tasks [55].

Beyond collaboration, additional dimensions of activity have been proposed in the literature and are further examined in this study. Mateescu [13], for instance, highlights attributes such as plannability, formality, complexity, and duration. These dimensions help reveal how work unfolds in time and space and how the environment can better support such processes. It is also important to note that even activities not formally scheduled in advance may still be intentional—studies show that 72% of unscheduled interactions are actually intentional, though not communicated in advance [13]. This challenges assumptions about the spontaneity of workplace interactions and underscores the importance of accommodating both planned and emergent behaviors in office settings. Another aspect is “complexity”, reflecting the cognitive and coordinative effort required, which can influence environmental needs, especially for tasks that require deep focus [15]. Duration varies widely—from short informal conversations to extended collaborative sessions—and is affected by spatial configurations and whether encounters are planned or spontaneous [14].

5.2. Assigning Activity Subclasses and Their Attributes

After compiling activity descriptions from the reviewed literature and the observational study, which included on-site observations, diary logs, and interviews, each activity subclass was assigned a corresponding set of attributes, forming a foundational profile of characteristics. Certain attributes, such as duration, modality, and number of participants (in the case of collective activities), required direct input from occupants during the activity scheduling process.

The definition of these attributes stems from the key information required for digital-twin operation, particularly regarding architectural and environmental adaptations. Some of the reviewed articles, namely Mokhtarian et al. [20], Duffy and Powell [50], and Hascher et al. [51], also classify activities according to their main defining attributes, providing a useful baseline for our framework. The attribute of “Social interaction” describes whether an activity involves one or multiple occupants, thus indicating the need for a more private or more collaborative environment. “Plannability” specifies whether an activity can be anticipated, influencing the type of adaptation to be implemented; when an activity is not predictable, the system must rely on sensors to detect the environmental stimuli affecting workers. “Posture” indicates whether the activity requires physical support or a specific stance, helping define spatial or furniture needs. Finally, “Complexity” refers to the mental effort and concentration required, which may imply a need for more controlled or private work settings.

Table 3. Baseline attributes assignment to activity subclasses by category.

General Classification	Subclass	Social Interaction	Plannability	Posture	Complexity
Focused Work	Coding	Individual	Schedulable	Seated	High
	Writing	Individual	Schedulable	Seated	High
	Reading	Individual	Schedulable	Seated	High
	Focused Individual Tasks	Individual	Schedulable	Seated	High
	Visual–Cognitive Tasks	Individual	Schedulable	Seated	High
	Cognitive–Administrative Work	Individual	Schedulable	Seated	High
	Research-Oriented Tasks	Individual	Schedulable	Seated	High
	Review-Oriented Tasks	Individual	Schedulable	Seated	High
Shallow work	Routine Office Tasks	Individual	Spontaneous	Seated	Low
	Physically Supported Office Work	Individual	Spontaneous	Flexible	Low
	Digital-Concentration Office Tasks	Individual	Schedulable	Seated	Low
	Setup and Closing Tasks	Individual	Spontaneous	Flexible	Low
Meetings—In-person/Online/Virtual meeting	Collaborative and Coordinative Meetings	Collective	Schedulable	Seated	High
	Informal Collaborative Activities	Collective	Schedulable	Flexible	Moderate
	Strategic Collaborative Work	Collective	Schedulable	Seated	High
	Spontaneous Collaborative Interactions	Collective	Spontaneous	Flexible	Moderate
	Presentation Activities	Collective	Schedulable	Flexible	Moderate
	Training and Learning Sessions	Collective	Schedulable	Seated	Moderate
	Brief Telecommunication Work Interactions	Individual	Spontaneous	Flexible	Moderate
Resting	Social Interactions and Breaks	Collective	Spontaneous	Flexible	Low
	Brief Telecommunication Breaks	Individual	Spontaneous	Flexible	Low
	Occupational Physical Activity	Individual	Spontaneous	Flexible	Low
	Individual Break	Individual	Spontaneous	Flexible	Low

shows the baseline assignment of attributes to the classes and subclasses of office activities.

The baseline attributes assigned to each activity class and subclass serve as a foundation for the schedule database, which also receives direct input from occupants when they schedule an activity. Once the activity is scheduled, occupants are prompted to answer five key questions regarding the number of participants, location, duration, modality, and any necessary modifications to the baseline activity attributes. These responses help refine the characterization of each activity, supporting the decision-making process for the Building Management System (BMS) to activate appropriate adaptive layers and environmental controls. In parallel, spontaneous activities, those not scheduled in advance, may be identified in real-time through automated systems within the building. These unscheduled activities are assigned to a lower priority, as their unpredictability limits the building’s ability to prepare in advance. Figure 7 illustrates the decision-making framework that supports this adaptive management approach.

A key gap in current activity-recognition approaches is the tendency to focus on classification accuracy or sensor-based detection but offer limited guidance on how activities should be described, parameterized, and operationalized for building control. As a result, BMS and digital twins often lack a standardized, actionable way to translate “what people are doing” into “what the building should do”, to then support decision-making for adaptive architecture. The taxonomy proposed in this paper addresses this gap by defining office activities together with a set of baseline attributes, i.e., participants, duration, spatial configuration, modality, and environmental requirements, that can be identified by both scheduled and spontaneously recognized activities. By embedding these descriptors in the operational framework, the taxonomy establishes a common language that connects occupant input, automated recognition, and adaptive architectural responses, offering a structured foundation for more anticipatory and context-aware building management in future hybrid workplaces.

5.3. Key Stressors in Office Environments and Their Impact on Everyday Work Activities

Workplace stressors affect the well-being and work performance of employees and are also the cause of adaptive behaviors that employees perform in the workplace. They wear headphones due to noise, for example, or they open a window and release air due to bad air quality. All these activities form part of their work and at times even break up tasks like focused work or attending meetings. Through these coping activities, the employees attempt to become as comfortable as possible at the workplace. These repeated adjustments reveal that current office environments often fail to provide conditions that adequately support the activities being performed, placing the burden of environmental regulation on workers themselves.

During the observational study, occupants reported performing adaptive behaviors themselves when they perceived disruptive environmental conditions such as increased noise, lack of privacy, uncomfortable temperatures, prolonged sitting, and glare. The most frequently mentioned adaptive behavior observed by researchers was the opening of windows to ventilate the room because of bad air quality. The second most common adaptive behavior was the use of noise-canceling headphones as an adaptation to high office noise caused by either chatting or specific work activities. Some of the participants wore headphones only to block office noise, without playing any sounds, such as, for example, music. These individuals were mostly engaged in focused work. Some participants who made phone calls during the observation period left the office to do so, in order not to disturb their colleagues with noise, which can be seen as yet another form of adaptive behavior.

Importantly, these behaviors illustrate how workers continuously negotiate their relationship with the office environment, with frustration, cognitive overload, and loss of focus emerging when the space fails to support specific activities and personal needs, while intelligent, situationally aware systems that assume some of the regulatory work of adjusting environmental conditions can alleviate this emotional and cognitive burden and contribute to more sustainable wellbeing over the working day.

Specific considerations about each environmental stressor domain are discussed in the following paragraphs.

• Noise

In general, the most widespread environmental stressor in shared or open-plan offices is noise [56,57,58], and this is similarly reflected in the fieldwork data from Leuven and Perugia. In the former, participants frequently mentioned the presence of noise, especially in Office B. Office A was considered the quieter office. The noise they pointed out was mostly related to people talking, or conversations within the offices, as illustrated by a quote from one of the participants describing the conversation noise in the office: “Some people… don’t really seem to consider that there’s others. Then I can get quite annoyed”.

They were also disturbed by conversations coming from the meeting room adjacent to Office A. Some even mentioned being bothered by voices from the hallway or outside, which are particularly audible when the office windows are open. In addition to conversations, devices present in the offices also contribute to noise. Participants specifically mentioned the noise of the printer. A few also referred to the vibration of mobile phones—their own and their colleagues’—which also disrupted their workflow.

In the Perugia case study, several interviewees reported being noticeably disturbed by conversations among office mates, particularly during physical or online meetings, which were almost always held in the shared offices. Additional noise-related disturbances included colleagues chatting in nearby corridors and construction work occurring during the interview period. Rather than asking others to modify their behavior, participants in Perugia generally adopted individual coping strategies, such as the use of noise-canceling headphones. One participant confirmed that by saying, “I cannot do hard work if I do not wear noise cancelling headphones.” During the winter months, windows were typically kept closed to maintain thermal comfort, which also reduced the intrusion of outdoor noise. However, this approach potentially impacted indoor air quality, exemplifying a multi-domain behavioral effect, where actions aimed at addressing environmental factors referring to specific domains (acoustics and thermal) inadvertently influence another (air quality).

Another way participants handled noise was to move to a different space when the office became too loud. The biggest problem highlighted was the lack of alternative spaces available for focused work or attending online meetings in both study cases. Others rescheduled their tasks based on the noise levels in the office. Knowing that noise would normally die down early in the morning and in the later part of the day, occupants chose to perform cognitively demanding tasks at that time. Others even worked in the office late into the evening or visited on weekends when there was no one around, and they could work without distractions.

Such behavioral strategies signal a misalignment between environmental conditions and the cognitive or social requirements of specific activities. A situationally aware building would instead detect the ongoing activity (such as focused work requiring low noise levels) and autonomously regulate acoustic conditions through adaptive panels, zoning, or notifications before the discomfort arises. This reduces the emotional strain associated with constant self-regulation and supports uninterrupted task performance.

While the majority of the literature [56,57,58] points towards noise as the main source of stress, our interviewees were not always content without noise per se. On occasion, they even complained that Office A was too quiet. One of them likened it to a library and explained how, on those quiet days, they would not unzip their backpack for fear of breaking people’s concentration with the noise of the zip.

An illustration of a positive experience of noise is demonstrated in participants’ comments about feeling more a part of the working environment through listening to what other people have to say. For example, one of the employees mentioned: “It’s part of this curiosity… if I hear something I kind of want to know what’s going on”. Another employee enjoyed working in a location where something is happening: “I like to work where stuff is going on”. Thus, we can say that occasionally noise can also make colleagues feel a sense of togetherness and reduce loneliness. To this extent, noise has an ambivalent character. These findings further emphasize the necessity of situationally aware systems that are able to make distinctions between productive ambient noise and disrupting noise, thereby allowing the building to fine-tune its response rather than relying on static acoustic strategies.

• Air quality

Although no sensor-based air quality measurements were conducted as part of the study, participants still spoke at length about the topic in Leuven. They understood air quality as a multi-sensory experience and associated it with both smell and temperature in the offices. Typically, smell was the more prominent concern, while temperature was mentioned in the context of window opening. In Perugia, air quality did not emerge as a major concern among participants, as it was rarely mentioned during the interviews. The offices generally offered good ventilation, and it was common practice for occupants to open the windows upon arrival in the morning to refresh the indoor air after a night of the windows being closed. Additionally, participants had access to “tilt-and-turn” windows, allowing for adjustable ventilation throughout the day.

Ref. [59] shows that indoor air pollution arises due to a lack of ventilation, human activity, and the presence of various materials, chemicals, and gases. In Leuven, participants themselves identified poor ventilation and its consequences on air quality as a concern: “For the office, one of the biggest complaints is about ventilation”. They also directly linked poor air quality to human presence and activity in the room: “It smells like people there. And there’s no ventilation”.

They reported bad smells and stuffy air in the offices, sometimes to the extent that it was hard to breathe: “Sometimes it smells bad inside of that room […] sometimes it’s unbreathable there.” For a minority of participants, poor ventilation was an even bigger issue than office noise: “But the air quality and the heat is just so bad that yeah, you it’s really, really annoying. I think it’s the main problem in in the office.” Very strong words were employed by participants to articulate air quality, like “unbreathable,” “really annoying,” or “very uncomfortable,” and these are signs of air quality having a strong influence on participants’ wellbeing at work. In fact, bad air quality has a dramatic influence on employee well-being and intellectual functioning [60].

The means by which participants avoided air-quality-related issues included leaving windows open—during our study, this was a challenging task, considering the colder conditions outdoors; leaving the main door to the office open (thus allowing noise in the hallway to spill in as well); or relocating to another office. However, this too was constrained by the availability of other rooms. Most participants believed that many of the air quality issues could be resolved through proper ventilation in the offices. In a situationally aware building, ventilation would automatically adjust for occupation levels and types of detected activities, which would reduce the need for workers to interrupt work to perform a physical intervention. This constitutes a good example of how environmental discomfort is not only physical but also an emotional and cognitive burden that intelligent systems can contribute to alleviating.

• Temperature

Indoor temperature significantly affects both the thermal comfort of employees and the perceived air quality in the workspace, as already mentioned in the previous subsection, as well as the performance of employees at their work [61,62,63]. Similar patterns were observed in our study. One participant specifically highlighted their sensitivity to temperature during the interview: “Another thing which I think affects me… the working environment in the office is the temperature.”

During the research period, dissatisfaction emerged among participants due to low indoor temperatures, which were largely the result of open windows: “In the winter… I don’t like the windows being open.” Employees in Leuven usually opened the windows to alleviate the poor air quality in the offices. In some cases, they asked their colleagues for permission, but in most cases they did not—which caused frustration among others: “I have a colleague… who always opens the window.” Some participants ended up working in their jackets because of the cold: “Because I’m not really good at dealing with the cold. So, if people put… if people open the window. I’m the first one to get cold usually and it really annoys me that when I have to sit there with my jacket on while working”.

In Perugia, similar conflicts arose due to differing thermal sensitivities among individuals sharing the same office. Some participants expressed discomfort when colleagues kept the windows open despite low outdoor temperatures, creating tension within the workspace. Conversely, others reported discomfort when the HVAC system was set to maximum cooling, leading to feelings of being cold. However, the availability of a well-equipped set of Personal Environmental Control Systems (PECS), including individual heaters and fans, helped occupants to partially mitigate these disagreements by allowing occupants to adjust their immediate surroundings according to personal preferences.

There were also reported issues in raising the issue of opening the window since nobody wanted to come across as too much of a “complainer”. As a result, the participants would often refrain from doing anything and instead suffer through the discomfort out of fear of offending their co-workers. One interviewee explained it like this: “And then everyone… You know that everyone feels it. So, I don’t dare to close or open the window sometimes because I know other people that are having the same discomfort”. Participants also mentioned occasional problems with heating, which further contributed to thermal discomfort in the office: “Otherwise, during the during the weekdays, if there is some issue with the heating then also if it gets colder inside the office”.

The negotiations around temperature control illustrate how environmental discomfort becomes a social and emotional issue, not only a technical one. A situationally aware building would integrate user preferences, occupancy distributions, and activity types to regulate thermal conditions automatically, thereby reducing interpersonal tensions and preventing comfort-related task disruption.

• Glare

Glare negatively impacts workers in offices, as it causes sensations of annoyance or even pain, which decreases their job satisfaction and, of course, their productivity [64]. Participants mentioned glare as one of the many distractions in the office: “Especially in this office, in the desk facing that way, you get the sun, let’s say on your back and you get a lot of glaring in your screens.” Glare mostly disrupts work, making it difficult to concentrate: “But the only thing that sometimes bothers me is the sunlight that comes directly to my eyes or to my monitor, depending on which side, I think. You just close the blinds.” As indicated in this quote, participants cope with glare by lowering the blinds.

Greater discomfort was reported by those seated either facing or with their backs to the windows—either the sunlight shines directly into their eyes if they are facing the window, or it reflects on their screens if their backs are turned to it. Both offices are equipped with automatic blinds that employees use as needed. Regarding lowering the blinds, participants were much less hesitant than with opening windows, possibly due to the fact that blinds impact fewer individuals around them. Perhaps it is because even when all blinds are closed, the office lights are usually already lit anyway, so people still receive sufficient light in which to work without glare.

In a situationally aware building, glare management would be automated based on sensor inputs and the activity being performed, for example, by dimming lights or adjusting blinds during visual–cognitive tasks to prevent the visual fatigue and emotional irritation associated with harsh lighting conditions. This again illustrates the link between environmental stressors, emotional responses, and the need for intelligent, activity-aware control.

6. Conclusions

Understanding the nature of activities performed in office environments offers critical insights into advancing human-centric Building Management Systems (BMS). Such systems are indeed expected to recognize occupants’ needs and dynamically adjust indoor conditions to support performance, health, and overall well-being, while also supporting personalization based on individual preferences and activity requirements.

In this study, a taxonomy of activities in office spaces was developed through a two-phase process: a comprehensive literature review of key studies related to office work activities and its validation through two case studies: one conducted in small, shared offices and the other in an open-plan workspace. Both involved occupant interviews and self-reported data on activities and comfort, while the open-plan case also incorporated direct observational research to capture occupant behaviors, interactions, and spatial preferences in real time.

The systematic review supported the identification of distinct classes and their respective subclasses of activities, which were subsequently refined and validated through the empirical data gathered during the case studies. Four primary activity classes emerged, namely Focused Work, Meetings, Shallow Work, and Resting, each one comprising various subclasses derived from common characteristics observed in both the reviewed literature and occupant-reported tasks. A baseline characterization of each subclass was then established, using key raised dimensions: social interaction, plannability, posture, and task complexity. Within a prospective Digital Twin framework integrated with the BMS, occupants could actively register their planned activities, modify default attributes, and input additional contextual information, such as duration, number of participants, modality, and location. This enriched activity profile would enable the system to anticipate environmental needs, autonomously configure room conditions for scheduled tasks, and respond dynamically to spontaneous interactions detected by the sensing infrastructure.

Although the empirical component of this study was based on a relatively small qualitative sample, the combination of observations, diary logs, interviews, and insights from the literature review provided a robust foundation for defining and validating the proposed activity taxonomy. By integrating these findings into a standardized activity vocabulary, the study demonstrates how such a framework can support context-aware Building Management Systems and Digital Twin applications in recognizing and responding to occupants’ needs.

This taxonomy also aligns with broader policy frameworks, particularly the Sustainable Development Goals (SDGs). By enabling a more precise and operational classification of office activities, the framework supports SDG 7 (Affordable and Clean Energy), SDG 11 (Sustainable Cities and Communities), and SDG 12 (Responsible Consumption and Production) through more efficient and adaptive building operations. Usanova et al. [65] showed that integrating SDGs into business models for the built environment fosters innovation, circularity, and resilient urban development. This taxonomy can also provide the informational foundation required for buildings to operate as intelligent, sustainability-oriented systems capable of anticipating and responding to occupants’ needs.

Future research should extend validation efforts to a broader range of workspace types, including home offices, hybrid settings, and coworking environments, and involve more diverse participant groups, enabling further refinement of the taxonomy across organizational and cultural contexts. In addition, incorporating user feedback on perceived environmental adaptations will be essential to improve system responsiveness and enhance the accuracy of activity-recognition models.