An Alternative Approach for Sustainable Management of Historic Urban Landscapes Through ANT via Algorithms: The Case of Bey’s Complex Palace in Constantine, Algeria

Abstract

Historic urban landscapes, despite their cultural significance, often face neglect, limiting their potential to increase the value of historical centers. Defined as a complex sociotechnical network that involves a variety of agencies incorporating material, immaterial, natural, and artificial elements, these landscapes present significant challenges for architects because of their layered and diverse components. Actor–network theory (ANT) is used as a methodological and ontological framework to address this complexity. However, a notable research gap exists on the basis of the lack of clear representation and practical application of ANT to address the complexity of these historic urban landscapes. To bridge this gap, this study uses Bey’s palace as a case study to develop a comprehensive framework based on a digital mapping approach rooted in ANT. This framework traces, visualizes, and analyzes historic urban landscapes as intricate systems of agencies, leveraging graph theoretical algorithms and computational analysis tasks from network analysis tools to increase their effectiveness. This investigation is based on two key concepts: the actor/actant and the actor network. The research employed Bruno Latour’s concepts of translation, agency, and the mapping controversies technique grounded in graph-theoretic algorithm tasks to decipher the complexities of Bey’s palace system. The results identify seven clusters as actor networks and highlight the roles of key actors/actants, such as Ahmed Bey, decorative elements, courtyard gardens, and Moorish architecture. This methodological approach provides architects and urban planners with practical tools to better understand, analyze and preserve historic urban landscapes, enriching their cultural and historical value. By transforming contested discourses into measurable networks indicators, this interdisciplinary framework directly supports SDG11 (Sustainable Cities and Communities), especially Target 11.4, in safeguarding cultural heritage by enabling the prioritization, monitoring and governance of cultural, social and infrastructural assets in historic urban landscapes.

1. Introduction

UNESCO has highlighted the preservation and management of historic urban landscape HUL as critical challenges for sustainable urban development [1]. These HULs serve as vital repositories of a city’s cultural, social, and architectural heritage [2]. They reflect the dynamic interactions between human and nonhuman elements over time [3]. These landscapes encompass both tangible elements, such as buildings, open spaces, and urban infrastructure, as well as associated ecological systems. Additionally, they include intangible aspects such as traditions, social practices [4], and cultural values linked to specific locations [5].

Despite their importance in preserving collective memory and improving cultural identity, HULs usually face neglect due to the complexity of their diverse and interconnected elements [6]. This complexity presents considerable challenges for architects, urban planners, and heritage professionals in their endeavors to manage and preserve these invaluable cultural assets. The diverse nature of the historic urban landscape requires the development of methodological and ontological frameworks that can effectively address its heterogeneous composition and dynamic associations. Accordingly, ANT, as articulated by Callon [7], Latour [8], and Law [9], provides a robust framework for understanding and managing historic urban landscapes. It emphasizes the interdependencies between human and nonhuman actors in shaping urban realities [10]. Numerous studies have directly applied ANT to examine HULs, offering a distinctive perspective on the interactions between human and nonhuman entities in the formation of historic urban environments [11].

However, while ANT provides a robust theoretical basis, its practical application in HULs studies remains underexplored. The existing research lacks comprehensive methodologies and tools to systematically visualize and interpret the intricate actor networks that define urban landscapes in historic centers, leaving a critical gap in theory and practice.

This research focuses on the following research question: How can ANT be systematically applied to define, visualize, and analyze the complexities of historic urban landscape HUL as a network of agencies, and what are the most effective tools for achieving this goal? This study has focused on the hypothesis that integrating ANT’s translation process with graph-theoretical algorithms offers a more effective framework for visualizing and analyzing the complexities of HUL than traditional methods do. Additionally, it posits that the cartography of controversial techniques uncovers hidden relationships and dynamics among both human and nonhuman actors, and that the proposed methodology enhances the ability of architects and urban designers to preserve and manage historic urban landscapes.

This study concentrates on the historic urban landscape of Bey’s Palace Complex in Constantine, Algeria, a site celebrated for its Moorish architecture, intricate decorative elements, and rich historical layers. Accordingly, this research aims to demonstrate how ANT, when paired with graph-theoretical algorithms from network analysis software, can reveal the complexities of historic urban landscapes by revealing the agency of key actors that shape these environments. Furthermore, the study intends to employ ANT to systematically and ontologically trace, visualize, and analyze the relationships between human and nonhuman components within historic urban landscapes, framing them as complex systems characterized by diverse and heterogeneous actors.

This study makes a substantial contribution to urban studies, architectural theory, and heritage management. It presents methodological innovations by introducing an integrated ANT framework, along with visualization tools based on various algorithmic tasks, designed to map intricate actor networks. An in-depth case study of Bey’s Palace Complex offers valuable insights into the cultural significance of the historic urban landscape. The findings yield actionable recommendations for the management of HUL, aiming to preserve them as vibrant cultural assets for future generations.

Moreover, ANT should be linked to sustainability. The historic urban landscape approach centers both material and nonmaterial values and is therefore inherently connected to sustainability questions about the cultural persistence, social wellbeing and governance of urban form. By operationalizing ANT into measurable network indicators, this study provides tools that can contribute to sustainability assessment, for example, by identifying which physical elements, practices or actors are central to maintaining social value; by flagging fragile clusters that may indicate vulnerability to socioeconomic change; and by detecting key factors such as obligatory passage points that function as governance chokepoints for policy intervention. These capabilities make this proposed framework directly relevant to sustainability-oriented heritage management and planning.

To achieve these objectives, this study explores the black box of the urban historic landscape surrounding Bey’s complex palace through visual translation. By utilizing the cartography of controversies technique [12] as a practical tool within Bruno Latour’s ANT translation process [13], as well as his concepts of structure and agency, this research examines the various components involved. It draws from digital actor network mapping as a relational sociotechnical method, employing multiple graph-theoretic algorithms and computational analysis tasks from network analysis software.

The research is structured as follows: the first section provides a detailed database on ANT and its interpretative methods regarding the elements of spatial agency within urban landscapes. It also reviews previous studies to establish the vocabulary of the theoretical framework. This research explores the mapping controversy technique as a valuable methodological and ontological tool rooted in graph-theoretic algorithms. The practical case study focuses on the urban landscape of the historic Bey complex in Constantine, Algeria, demonstrating how the concepts of spatial agency components and actor networking were developed through a virtual tracing of their presence. This analysis illustrates how these elements enhance the cultural significance of the historic complex in Constantine and concludes with key recommendations and insights. Figure 1 outlines the workflow process of the research study.

2. Literature Review

2.1. ANT’s Historical Background

ANT emerged in the 1980s as a key component of the broader field of science and technology studies (STS), with key contributors such as Bruno Latour, Michel Callon, and John Law. Their research focused on understanding how social, scientific truths, and technological artifacts are constructed [9,14], which is particularly useful for analyzing challenges in urban planning and design [15].

ANT is considered one of the most significant approaches grounded in rationality, fostering fragmented projects that engage with one another to create a cohesive whole. ANT emphasizes the integration of heterogeneous agents, encompassing natural, material, and cultural elements of the urban environment along with human actors [8]. By rejecting the principle of dualities that separates various differences, ANT promotes the formation of hybrid groups grounded in concepts such as power, authority, and representation [16].

ANT fundamentally revolves around the concepts of actor/actant and actor networks in the interpretation of complex systems. According to Latour [14,17], this perspective is crucial for understanding the interconnections among various entities.

Callon [7] highlights that a key characteristic of ANT is its view of space not only as a perceived entity but also as a construct arising from the interactions and performances of both human and nonhuman actors. This framework treats people and objects as equal agents that are interconnected within networks that are created and maintained to fulfill specific purposes.

2.2. ANT View of the Urban Landscape

Urban landscapes are increasingly recognized as dynamic and contingent spaces shaped by complex interactions among a variety of actors, including natural elements, infrastructure, governance systems, social practices, and historical influences [18]. ANT presents a distinct perspective on the interpretation of urban landscapes and their treatment within historic centers. ANT presents a balanced ontology that dismisses hierarchical distinctions, focusing instead on tracing interactions and their impacts across both space and time [19]. This approach enables us to examine urban phenomena as emergent properties of intricate networks, providing a more nuanced understanding of the processes underlying spatial transformation.

Ontologically, ANT’s focus on nonhuman agency challenges conventional human-centric narratives surrounding urban development. By acknowledging the influence of materials, technologies, and natural elements, ANT reinterprets the urban landscape as an assemblage with distributed agency [20]. This viewpoint resonates with contemporary urban studies, which increasingly explore hybrid landscapes where cultural and ecological dimensions are interconnected [21].

2.3. The Spatial Agency Component

According to specialized studies, ANT posits the existence of actors and actants within the urban landscape that form networks. These networks are characterized more by their heterogeneity than by being discrete entities. Bruno Latour articulates that agency, whether it is an actor or actant, is anything that alters the state of affairs by making a difference. This agency is recognized by others, as it can serve as either the subject or the objective of another activity. Theorists have differentiated agency on the basis of its mode of existence, categorizing it as either an actor/actant or an actor network, as follows:

2.3.1. The Actor/Actant as an Agency

Theorists specializing in ANT have focused on the actor/actant as the fundamental unit from which actor networks are constructed. Their definitions and classifications related to space are outlined as follows:

Latour [17] defines the actor/actant as any entity that is represented. In contrast, Ruming [22] describes actors/actants as entities that perform actions, encompassing both human and nonhuman agencies, mental capabilities, or crafted objects. He emphasized that the actor/actant plays a crucial role within the network or space, aligning with the overarching vision of the network. Each actor/actant, whether singularly or collectively, contributes to one or multiple networks or may even be a product of such networks.

Boelens [15] identifies a category of actors or actants that he terms principal agency, which the theory addresses in the development process. This type of agency possesses distinctive characteristics, the most significant being strength, the ability to take action, control associations, and the ability to motivate other actors to act in accordance with its traits. Additionally, it embodies exclusivity and multiple values. Conversely, another form of agency, referred to as organizational forms of the network [23], is defined as any group of actors aiming to establish frequent and enduring exchange relationships, although it lacks the authority to resolve disputes that may arise during these exchanges.

Concerning the types of urban landscape agency actors/actants, Allen [24] has shown that the agency (actor/actant) comprises humans and nonhumans and that the human agency (actor) can be (an individual with authority or a stakeholder, space manager, space worker, or space user) or that the nonhuman agency (actant) includes important agencies, whether material or immaterial (objects, concepts, texts, architectural or artistic inscriptions, techniques),or natural (plants, topographical features, animals, water) or intangible (identity, concept, cultural, event, economic, social or environmental value, collective memory).

Similarly, Erntson [25] was interested in the actor as a human being and categorized him into types (the users or groups of users and the holders of power, whether he or she is the owner, organizations, companies, department heads, or planners), counting urban landscapes with their physical and social components as influential and decisive agencies in achieving broad support for sustainable urban development, as they act as physical places of social-ecological interaction that can feed value creation processes and environmental knowledge to improve urban processes.

2.3.2. Actor Network as Agencies

Studies refer to the actor network as a group of agencies linked to relationships through different processes, and the interpretation of actor-network theorists is as follows:

Studies refer to the actor-network as a group of agencies linked by relationships and processes; actor-network theorists interpret these phenomena in terms of translation, enrollment and network effects. Several classic ANT studies emphasize how power and authority often emerge as properties of the network itself: networks create durable arrangements through the enrollment and stabilization of heterogeneous elements [7,8,9]. Works applying ANT to spatial and heritage contexts demonstrate how such network effects allow particular actors or configurations to consolidate others’ positions through specific practices and thereby connect and shape distant spaces [26,27]. From this perspective the network operates as an active agent capable of reconfiguring its components, which ANT authors term hybrid collectives or heterogeneous ensembles and can appear in many forms. Concrete examples include clusters of historic buildings embedded within an urban fabric, as well as associations, institutions, organizations and social movements that act collectively; these hybrid agencies help explain how some actors exercise influence at a distance and how certain spatial configurations persist or change. Accordingly, the abovementioned studies treat the actor-network as a hybrid agency whose effects are both material and relational.

Cvetinovic et al. [18] described the actor network as interconnected assemblages that treat both humans and nonhumans symmetrically. This perspective allows for the distinction of actors and actants, emphasizing that “the representative” does not serve as the source of action or activity; rather, it is a dynamic target influenced by a diverse range of entities mobilized toward it [8]. ANT prioritizes the formation of relationships over the intrinsic qualities of actors and actants, focusing on how networks are formed through organizing agencies that connect groupings on the basis of their properties, goals, functions, activities, users, or time.

The previous discussion clearly illustrates that the concept of agency, whether considered an actor/actant or a component of an actor network, includes any entity in the surrounding environment, be it human or nonhuman, material or immaterial, that has the capacity or power to take action. These entities are influenced by their relationships, emphasizing the need to refine the ANT approach to better understand the construction of these relational dynamics for agencies. This idea is encapsulated in the notion of actor networking.

2.4. Previous Studies: ANT Application in Historical Urban Landscapes

The ANT presents a unique perspective on the interpretation of urban landscapes and their treatment within historic centers. According to ANT, preserving the various layers and spatial agencies of these areas by embracing the urban landscape itself is essential. The theory emphasizes the creation and development of effective networks that incorporate different agency components, whether they are actors/actants or broader networks, through the concept of the actor network.

This approach serves as a strategy for managing cultural monuments, whether they stem from built agents, natural elements, or various spatial contexts. Therefore, understanding the terminology and mechanisms associated with actor networking as a means to preserve and enhance diverse characteristics and promote sustainability is crucial. This approach is supported by a range of specialized studies within the framework of ANT, as follows:

Ruming [22], in a study titled “A new network direction in housing studies: the case of actor network theory”, explored ANT as a theoretical model in urban studies, emphasizing the integration of human and nonhuman actors within heterogeneous networks. The study introduced the concept of translation, where mediators adapt and transfer properties within spatial networks to ensure their continuity. This approach highlights the importance of valuable agencies and heterogeneous interconnections in sustaining spatial dynamics, particularly in historical urban centers.

Similarly, Ernstson [25], in a study titled “In Rhizomia: Actors, networks and resilience in urban landscapes”, examined urban landscapes through ANT, focusing on the resilience of social, ecological, and material agencies. The study emphasized creating balanced relationships among these elements and highlighted the importance of actor networking in framing and enhancing the value of built and natural historical agencies.

Allen [24], in a study titled “On actor-network theory and landscape”, introduced the ANT landscape model, which frames urban landscapes as material social constructs that influence the local environment and foster intimate urban settings. This study introduced the concept of synchronous heterogeneity to describe the dynamic and diverse spatial experiences shaped by memory, sensory engagement, and cultural backgrounds. By examining how spatial agency interacts with consumption patterns and user needs, Allen revealed how urban landscapes such as shopping malls, heritage centers, and lively streetscapes reflect these dynamics. Additionally, the translation of diverse materials into a range of activities in urban landscapes was shown to foster unexpected interactions, enriching spatial experiences and urban functionality.

Additionally, Ernstson and Sörlin [28], in a study titled “Weaving protective stories: connective practices to articulate holistic values in the Stockholm National Urban Park”, introduced the concept of interconnected conservation as the process of weaving historical narratives using natural and cultural agencies to mobilize public participation and enhance preservation efforts. The study categorizes these agencies into historical, biological, and cultural components, emphasizing their role in fostering collective memory, social welfare, and ecological reproduction, thus creating a comprehensive framework for urban conservation.

Tietjen [29], in a study titled “Translations: Experiments in landscape design education”, examined the transformative potential of ANT in creative urban landscapes. The study explored interventions that translate spatial agencies into innovative actor networks, demonstrating how such transformations affect urban structures and uses. By framing these interventions as opportunities for innovation, the study highlighted the role of ANT in reshaping urban landscapes.

Dobson [30], in a study titled “Urban translations: Regeneration through the lens of actor networking” utilized ANT to analyze urban regeneration processes, emphasizing the interdependence of environmental, social, and economic factors. The study introduced actor networking as a framework for sustainable change, focusing on the creation of specific activities and the integration of external and internal resources to enhance resilience within urban systems. By promoting feedback loops among diverse actors. Additionally, in this study various strategies for enhancing the sustainability of urban regeneration initiatives are identified.

Kim [31], in a study titled “Designing multiple urban spaces: an actor network theory analysis of the multiplicity and stability of public space”, investigated the foundational principles of ANT in urban landscapes, focusing on concepts such as plurality, change, and stability. This study used Kingston Market and Fortune Park Street in London as case studies, and the research revealed how material agents influence urban dynamics. While Kingston Market emphasized stability through structured interactions, Fortune Park Street demonstrated flexibility and diversity, highlighting the role of material agents in negotiating urban spaces.

Despite the contributions of these previous studies, several limitations remain. One major shortcoming is the lack of clear representations and practical tools for visualizing and analyzing the actor networks and key agencies within the historic urban landscape system, as noted in the works of Dobson [30] and Ruming [22]; however, these studies highlight the theoretical potential of ANT, which falls short in providing concrete methodologies. Additionally, the research conducted by Kim [31], Ernstson [25] and Tietjen [29] is largely dependent on qualitative descriptions, which ultimately lack the precision necessary for thorough analysis.

2.5. Mapping Controversies and Graph-Theoretic Algorithms as Practical Tools for the ANT Translation Process

On the one hand, the translation process in ANT describes how actors negotiate, adapt, and align their interests and actions within a network [32]. This process is frequently marked by controversies, as various actors possess competing interpretations of facts, values, or interests [7]. Consequently, mapping controversies serves as a practical tool for the ANT translation process, allowing for the visual representation of actors and their relationships. This approach aids in understanding the dynamics of alliances, power relationships, and the flow of information within the network [13].

On the other hand, algorithms represent a set of instructions aligned with ANT, focusing on computational efficiency, which facilitates the establishment and execution of numerous continuous tasks [33]. This approach enhances the visibility of actors and translations, effectively illuminating the black box of complex sociomaterial networks. Furthermore, the application of computational tools from network analysis and parametric modeling has significantly advanced the study of actor networks by enabling the visualization of relationships between human and nonhuman objects [34].

Similarly, Yaneva [12] illustrates how post parametric tools can effectively map the dynamics of controversies and actor interactions over time and space. In addition, Palmer [35] emphasized the importance of computational tools from social network analysis in revealing actor–network behaviors and connections. These computational methods enhance the application of ANT by providing dynamic and engaging representations of complex sociotechnical systems, thereby deepening our understanding of their spatial and temporal dimensions [12]. Furthermore, Arul et al. [36] underscore the importance of graph-theoretic algorithms in network analysis as powerful instruments for analyzing and understanding the intricate linkages and structures within complex systems.

Accordingly, the integration of mapping controversies with graph-theoretic algorithms from network science presents a robust approach for analyzing ANT translation processes [37]. Similarly, Solli et al. [38] emphasized that by conceptualizing controversies as graphs, researchers can leverage the quantitative capabilities of graph-theoretic algorithms as a mathematical framework. This approach enables more profound insights into the dynamics of these controversies.

These graph-theoretic algorithms offer more objective and quantitative analysis measures of network properties, including centrality, density, and clustering coefficients [39]. These measures can be utilized to identify key actors, influential subgroups, patterns of information dissemination, and the overall structure of the network [40]. In this context, the nodes in the graph represent actors, whereas the edges signify the relationships between them, such as alliances, conflicts, and information flows [37].

In recent years, actor–network theory (ANT) has been increasingly applied to heritage, urban studies and sustainability topics, with several recent contributions demonstrating ANT’s usefulness for analyzing sociotechnical transformations, spatial reconstruction, and digitally mediated controversies [41,42]. However, existing work shows a pattern: while conceptual and qualitative ANT applications have grown, explicit computational operationalizations that combine ANT’s cartography-of-controversies approach with graph theory from network analysis tools (e.g., VOSviewer) remain comparatively uncommon despite the efforts of some emerging studies such as [43,44], Foundational tools and methods for network analysis and mapping (such as VOSviewer 1.6.20) are available and are increasingly used in related fields, which enables algorithmic treatments of controversy-maps, but their integration with ANT as a formalized framework is still emerging [45]. This observation situates our contribution: instead of restating ANT’s conceptual claims, the present study proposes and demonstrates a transferable computational framework (data collection, preprocessing, network construction centrality/cluster analysis) that operationalizes ANT’s translation process for historic urban landscape analysis and sustainability-informed heritage management.

This study builds on these developments by pairing cartographies of controversy with multiple graph-theoretic algorithms from network science to operationalize ANT’s translation process for historic urban landscape analysis. This correlation is presented across three complementary tables.

Table 1. Hybrid mapping of ANT translation moments, roles, and computational implementations (Panels A–C).

A. ANT translation moments, roles, and related computational tasks
ANT Moment/Concept	Role of ANT Process/Concept	Related Algorithms and Computational Tasks
Problematization	Defines the research problem, identifies contentious issues, candidate actors/actants and the scope of the controversy [7].	Natural-language processing (NLP): tokenization, TF–IDF weighting, and entity extraction [46].
Interessement	Creates alignments and stabilizes relationships by enrolling allies, framing issues, and negotiating roles [7,8].	Association-strength normalization and co-occurrence mapping [47].
Enrollment	Organizes and consolidates roles, leading to the formation of cohesive groups or assemblies of actors [7].	Graph Layout Algorithms using Force-directed algorithms based on the LinLog Layout algorithm [48].
Mobilization	Stage where enrolled actors act collectively to implement decisions, solidify alliances, and produce material changes or coordinated outcomes [14].	Community detection algorithm based on Louvain Modularity Algorithm [49].
Boundary Objects/Mediators	Act as interfaces that enable translation and coordination across heterogeneous actor groups [50,51].	Fractional counting [52] and intercluster linkage computation based on computational analysis via co-occurrence analysis algorithm [44].
Spokespersons/Focal actors/Obligatory passage point (OPP)	Actors or entities that represent, speak for [7] or coordinate the interests of a group within the network [8].	Centrality analysis (eigenvector/betweenness) and total link strength measures [36].
B. Functions of algorithms and related tasks in network analysis tools
Algorithms and Computational Tasks	Function of Algorithms/Computational Analysis	Related Tasks in Network Analysis Tools (VOSviewer/RAWGraphs)
Natural language processing algorithms NLP (TF–IDF) [46].	Extracts and weighs textual terms representing actors and concepts [47].	Preprocessing using VOSviewer input (text-mining module) [46].
Association Strength Normalization algorithm [47].	Computes relational proximity between terms to reveal associative structures [47,53].	“Association strength normalization” option in VOSviewer [52].
Force-Directed (LinLog) Layout algorithm	Displays network geometry through attraction and repulsion forces [48].	“LinLog/Force Layout” parameter in VOSviewer or RAWGraphs [53].
Louvain Modularity algorithm	Detects cohesive clusters/communities based on modularity Q [49].	“Cluster Detection” tool through Modularity normalization in VOSviewer software which report modularity Q and number of clusters [53].
Fractional Counting	Allocates partial co-occurrence weight to each document contributor [52].	“Counting Method via Fractional” option in VOSviewer [53].
Centrality Measures algorithms (Eigenvector/Betweenness)	Identifies influential/bridging actors controlling information flow [35,54].	Exploitation of computational analysis results from VOSviewer software and Regression analysis in RAWGraphs tool [55].
Computational analysis via co-occurrence analysis algorithm.	This algorithmic task identifies actors with a high number of joint occurrences with other actors in the network [46]. This indicates their influence on other well-connected actors within the network.	“Actors Details through co-occurrence analysis” field in VOSviewer and graph visualization in RAWGraphs tool [55].
C. Algorithm mathematical formulas and mechanism descriptions
Related Algorithms and Computational Tasks	Mathematical Formula	Description of Mechanism
Term Extraction/TF–IDF (NLP)	According to Erkan [56]. ${\mathrm{IDF}}_i=\mathrm{l}\mathrm{o}\mathrm{g}\left({\displaystyle \frac{N}{n_i}}\right)$	Calculates how important a term/item i is in document j within the corpus N, weighting frequency by inverse document frequency to identify key actors [56].
Association Strength Normalization	According to van Eck and Waltman [47]. $W_{IJ}={\displaystyle \frac{C_{IJ}}{C_I{C}_j}}$	Normalizes co-occurrence counts by marginal frequencies to measure relative association strength between actors i and j [47].
Force-Directed (LinLog) Layout	according to Noack [57]. ${\displaystyle \sum _{\left\{U,V\right)}}\left(w_{u,v}{\displaystyle \frac{\parallel p_u-p_v{\parallel }^{\mathrm{a}+1}}{\mathrm{a}+1}}-w_u{w}_v{\displaystyle \frac{\parallel p_u-p_v{\parallel }^{r+1}}{r+1}}\right)$	Minimizes energy E to position nodes; attraction ∝ link weight w and repulsion ∝ distance d, forming visual clusters [57].
Louvain Modularity algorithm	According to Blondel et al. [58] $Q={\displaystyle \frac{1}{2m}}{\displaystyle \sum _{ij}}\left[A_{ij}-{\displaystyle \frac{k_i{k}_j}{2m}}\right]\mathsf{\delta }\left(C_I,C_J\right)$	Maximizes modularity Q to partition the network into communities with dense internal links and sparse external links [58].
Fractional Counting algorithm	According to Perianes-Rodriguez et al. [52] $W_{IJ}={\displaystyle \frac{C_{IJ}}{{\displaystyle {\sum }_K}{C}_{IK}}}$	Assigns fractional co-occurrence weight to balance contributions from multiterm documents, reducing bias from prolific items [52].
Eigenvector Centrality algorithm	According to Bihari and Pandia [54] Ax = λx	Computes node influence as proportional to the sum of neighbor influences; high-value nodes connect to other high-value nodes [54].
Co-occurrence analysis algorithm	According to Van Eck and Waltman [46] $c_{ij}={\displaystyle \frac{\mathrm{Count}\left(i,j\right)}{\mathrm{Total} \mathrm{Count}}}$	Computational tasks based on co-occurrence frequency calculation between all items appearing together, this allows to identify bridging or mediating actors [46].
Total Link Strength (TLS) analysis algorithm.	$TL{S}_i={\displaystyle \sum _j}{w}_{ij}$	Aggregates all link weights connected to node i, quantifying its overall connectivity and stability in the network.

—A summarizes the conceptual correspondence between ANT translation moments, their theoretical roles, and the related computational tasks. Table 1—B details the practical functions of these algorithms and how they are implemented in network-analysis environments such as VOSviewer 1.6.20 and RAWGraphs 2.0. Finally, Table 1—C outlines the mathematical mechanisms and equations underlying each computational process, demonstrating how quantitative measures—such as term weighting, association strength, modularity, and centrality-reflect ANT’s interpretive principles.

Table 1 provides a compact, three-panel translation guide that operationalizes ANT’s conceptual vocabulary into a reproducible computational pipeline. Panel A defines the theoretical moments and the roles or actor-types that the analysis seeks to identify; Panel B maps those conceptual tasks to concrete algorithms, analytic functions, and the specific software or visualization tools employed; and Panel C specifies the key mechanistic computations and formulae that produce the numeric indicators (weights, similarities, centralities, TLS, modularity scores) used in interpretation.

3. Materials and Methods

3.1. Criteria for Selecting the Case Study

Bey’s Palace (Constantine, Algeria) was selected because it presents a set of verifiable features that directly inform and enable the operationalized ANT pipeline used in this study. The case is described here only in terms of characteristics that serve as inputs or interpretation cues for the computational workflow: (1) built morphology articulated facades, courtyard systems and clustered fabric that produce place-specific toponyms and architectural terms used during text and archival searches; (2) documented discourses and controversies visible in local news, public reports and archival records that supply the textual corpus for controversy mapping; (3) the presence of multiple stakeholder groups (municipal authorities, heritage agencies, local residents and civil organizations) whose statements and documents appear in the collected sources; and (4) references to material and craft practices in source materials that function as likely boundary objects in network maps. These empirical features were the only descriptive elements used to guide term extraction, node selection, and the identification of candidate Obligatory Passage Points (OPPs) in the analysis.

The case is used to demonstrate the proposed framework’s steps (data collection, preprocessing, term extraction, network construction, cluster/centrality analysis, and interpretation) and to operationalize ANT translation moments (problematization, interessement, enrollment, mobilization and stabilization). This selection is methodological rather than statistically representative: the aim is to present a transferable computational approach that can be replicated and tested in other historic urban landscape contexts.

Additionally, this selection addresses a gap in current research, as few studies have systematically applied ANT to historic urban landscapes, particularly within the Algerian context. Focusing on this culturally significant site deepens the understanding of its historical and spatial dynamics while offering methodological innovations applicable to other historic urban landscapes. Ethical considerations are observed throughout, ensuring objectivity in data collection and respecting privacy and confidentiality.

3.2. Data Collection Steps, Sampling and Sources

Research data were collected to examine urban landscapes within historic centers, with a focus on the cultural significance of Bey’s palace in Constantine. A qualitative approach was employed to investigate controversies surrounding this landmark, utilizing network data analysis supported by the text-mining capabilities of VOSviewer software, which incorporates natural language processing algorithms (NLPAs) as described by Van Eck and Waltman [46]. This method enables tracing and identifying interactions among stakeholders by analyzing co-occurrences in textual data, ultimately generating a network of interconnected actors.

The data collection process comprised four primary stages, as adapted from the study conducted by Bukar et al. [59]. Figure 2 shows a comprehensive illustration of this process. Therefore, the workflow includes the processing of data from its sources to preliminary network visualization, as described below:

Stage 1: Data collection. This stage involves collecting information regarding the controversies surrounding Bey’s palace from diverse sources. This includes relevant documents found online, such as master’s theses, local social media posts, national journal articles, and onsite observations. Five pertinent videos were selected from YouTube through a careful screening process, ensuring that they featured expert interviews related to Bey’s palace. The transcripts of these videos, along with all accompanying data, were compiled into a single text file.

Stage 2: Data cleaning. The process of data cleaning is often a crucial step in creating network maps from textual data [46], which is essential for computational text analysis to extract meaningful insights. Initially, data preprocessing was carried out on the previous text data file to remove irrelevant information, such as punctuation, noise, and symbols. At this stage, VOSviewer accepts solely compiled, cleaned, and formatted network data in a text-based file for further analysis. As a result, the attributes of the dataset related to the case study were specifically modified to align with VOSviewer’s requirements.

Stage 3: Tokenization. Lexos software is utilized to improve data cleansing and transformation through scrubbing and tokenization processes. As part of an integrated lexomics workflow, Lexos is a web-based tool that assists users in exploring a chosen collection of digitized texts by employing computational and statistical techniques on the text corpus. This method encourages a more reflective and knowledgeable examination of the content [59]. Data scrubbing refers to the process of eliminating personally identifiable information from the text [60]. Tokenization refers to the division of lengthy text into smaller segments or tokens for analysis [61]. Lexos software was subsequently integrated to further enhance the VOSviewer text analysis.

Stage 4: Text analysis. The tokenized dataset is processed via VOSviewer software, which facilitates the automatic identification of the most prominent stakeholders associated with the controversies surrounding Bey’s palace. This analytical stage serves as a foundation for a more in-depth examination aimed at clarifying the translation process within ANT.

3.3. Procedural Workflow Process of the Proposed Method

In this analysis, a three-stage procedural approach is adopted to systematically deconstruct the controversies surrounding Bey’s palace while effectively applying the four moments of the translation process within ANT [7]. Grounded in the mapping controversies technique [12], this method allows the creation of analytical diagrams that illustrate a sociomaterial perspective, mapping the interactions between human actors and technical components. To address these controversies, precision digital tools featuring a variety of graph types and optimization algorithms are utilized. VOSviewer version 1.6.20 is employed to explore and analyze networks of scientific data [62], whereas RAWGraphs serves as an open-source tool specifically designed for visualizing complex data [55]. The procedural workflow is organized as follows:

The initial stage involves identifying and tracking the influence of various actors or controversies throughout the ANT transition process. This stage is crucial, as it allows for the organization of data within the software, utilizing a comprehensive collection of relevant information pertaining to the case study controversies. This process enables the generation of co-occurrence networks for the most significant actors through natural language processing algorithms in VOSviewer software [46]. This stage is similar to the moment of problematization in the ANT translation process.

In the second stage, various visualization techniques are employed on the basis of relevant algorithmic tasks to map and display the network layout of interconnected actants. The total link strength and co-occurrence between actors serve as weighting factors. Visualization techniques, such as network visualization and network density visualization of items [63], are executed through the force-directed algorithm [48], which is founded on association strength algorithmic analysis [53]. This stage reflects the moment of interessement within the ANT translation process, enabling us to visualize the intermediary actors within the network. Additionally, this stage offers an initial visualization of several key actors, highlighting the obligatory passage point OPP and the focal actor within the network structure of the case study. A more detailed discussion of these key actors in relation to the ANT translation process is presented in the results and discussion section of this paper.

The third stage comprises two consecutive steps. In the initial step, a cluster density visualization is utilized to identify the subsystems within Bey’s palace network structure. These subsystems represent clusters of actants that possess shared attributes. Each cluster is distinguished by a specific color range. The identification of these clusters relies on two complementary and synchronized algorithmic tasks, the LinLog layout algorithm and the Louvain modularity algorithm, both of which are implemented through the LinLog/Modularity analysis feature in VOSviewer version 1.6.20. This step is associated with two successive moments of the ANT’s translation process namely, the enrollment moment and the mobilization moment. A further explanation of these algorithmic mechanisms is provided in the results section.

In the second step, a regression analysis was performed via graph visualizations generated with the RAWGraphs tool [55]. This analysis relied on the outcomes of computational tasks performed in VOSviewer software and utilized two algorithms: the co-occurrence analysis algorithm and the eigenvector centrality algorithm. Additionally, it considers relevant parameters such as total links, total link strength, and co-occurrence frequency among actors within the network. This approach aimed to illuminate the relationships between actants and identify the key players influencing the dynamic network of Bey’s palace landscape. These pivotal actants align with several key concepts in the ANT translation process [64], including mediators, spokespersons, focal actors, and obligatory passage points. A more detailed explanation of these mechanisms from the second step can be found in the results and discussion section.

Additionally, the fractional counting algorithm is utilized through fractional optimization analysis in VOSviewer software [52,53] to identify boundary objects. The discussion will thoroughly cover these key actors in the results section. The steps of the proposed method are summarized in Figure 3.

4. Case Study Analysis

The empirical material that follows (Bey’s Palace, Constantine) is provided primarily to demonstrate and validate the operationalized ANT pipeline developed in this study. The palace is therefore treated as an analytically rich exemplar selected because its combination of tangible architectural features, contested meanings, archival sources and multiple stakeholder groups produces the kind of heterogeneity required to test how ANT translation moments (problematization, interessement, enrollment, mobilization and stabilization) can be captured with computational, graph-theoretic tasks. Below, each subsection presents descriptive and visual evidence from the palace and begins with a short note indicating which ANT translation stage the empirical material is used to illuminate and how it feeds into the algorithmic steps reported in Section 3 and Table 1—(A–C). Figure 4 Shows Bey’s palace location.

4.1. Architectural Features of Bey’s Palace Complex

Illustrating the problematization moment, this subsection documents the palace’s built features and material articulations to identify the initial problem-space and candidate actants/actors that informed our textual and archival searches.

The palace features traditional Islamic architectural elements and reflects the organization typical of Ottoman-era palaces, emphasizing a combination of public, semiprivate, and private spaces. The architecture demonstrates a strong relationship between interior spaces and courtyards, following traditional Islamic principles of spatial hierarchy and privacy. As illustrated in Figure 5.

4.2. Architectural Aesthetics and Functionality of Bey’s Palace

Illuminating the interessement moment, the material and aesthetic features described here reveal affordances, frames and alignments that attract certain actor coalitions; we use these observations to explain patterns of association and co-occurrence in the textual corpus.

The patterned tiles surrounding the doors highlight motifs that beautifully merge Algerian and Moorish aesthetics, complemented by intricately carved cedarwood. The pointed arch serves as a transitional element, bridging the external public realm with the palace’s internal ceremonial spaces. Additionally, various frescoes embellish the walls around the garden courtyard, symbolizing Bey’s journeys and the cities he visited throughout the world. As illustrated in Figure 6.

Additionally, within the palace, the stained-glass windows create an interesting interplay of light and shadow. The vibrant patterns cast intricate designs onto the tiled floors. These windows serve not only as cultural and environmental mediators but also, beyond their aesthetic appeal, fulfill a functional role by maximizing daylight while maintaining privacy, which is an essential characteristic of Ottoman domestic architecture. As illustrated in Figure 7.

4.3. Courtyards Nodes of Social and Environmental Interaction

Demonstrating Enrollment and Boundary Objects, the courtyard spaces function as boundary objects and loci of enrollment in the local actor network; their spatial and social roles guide our interpretation of cluster formation and mediating items in the network maps.

The courtyards of the palace function as essential hubs for social interaction and environmental engagement, promoting community participation and enhancing thermal comfort. By combining vibrant landscaping with intricate paving designs, these courtyards exemplify the effective integration of natural and built environments, highlighting the influence of the Moorish design. As illustrated in Figure 8.

4.4. Materials and Traditional Crafts

In support of mediators and spokespersons, descriptions of materials and craft practices identify mediating elements and spokesperson claims that appear as bridging items or high-centrality nodes in our network analyses.

Bey’s palace exemplifies the synthesis of local craftsmanship and the incorporation of luxurious materials acquired from various sources. Notably, marble sourced from Italy and cedar wood procured from the local tribes of the Aurès and Kabylie regions are utilized, highlighting a distinct blend of local and international influences, as illustrated in Figure 9.

4.5. Adaptive Reuse in Postcolonial Algeria

For the mobilization and stabilization moment, the adaptive reuse trajectory is presented here to show how the enrolled actors and OPPs translate into coordinated action and longer-term stabilization (or instability) of the network observations that we map quantitatively in the mobilization/stabilization analyses reported below.

The palace has been transformed into a cultural heritage museum within postcolonial contexts. This transformation highlights the importance of historical spaces in shaping national identity and collective memory following colonial rule. The palace evolves into a venue for both celebration and daily life, with gardens, courtyards, and interior spaces nurturing interactions among rulers, citizens, and visitors. As illustrated in Figure 10.

Finally, the descriptive and visual material from Bey’s Palace has been presented not as an end in itself but as a series of affordances and controversies that validate specific steps in our computational ANT pipeline. Empirical observations from Section 4.1, Section 4.2, Section 4.3, Section 4.4 and Section 4.5 fed directly into our preprocessing and term-extraction choices, guided the interpretation of clusters and centrality measures, and helped identify candidate Obligatory Passage Points and mediators for follow-up analysis. In the Discussion (Section 5), we draw these method-driven findings together with the broader literature to show how the pipeline can be used in other historic urban landscapes and to discuss implications for sustainable heritage management.

5. Results and Discussion

This section presents the findings of the study and demonstrates the potential of the mapping controversies technique as a practical tool for applying ANT, which is supported by optimization tasks and computational analysis through network analysis software. The results are organized according to the four moments of the ANT translation process outlined in the Materials and Methods section.

5.1. Results of the First Stage

At this stage, initial data on the controversies surrounding Bey’s palace were collected and processed via VOSviewer, a network analysis software that depicts the structure of the case study. This corresponds to the problematization moment in Latour’s [8] translation process, which involves tracing actors and the influences shaping the definition of the problem.

The data were then formatted, processed, and organized through text mining functions in VOSviewer, which are supported by natural language processing algorithms (NLPAs). This computational approach involves extracting relevant terms and identifying the relationships among them [46].

Accordingly, the corpus described in Section 3 was transformed into a network of candidate actors through systematic preprocessing, and the term-selection framework below focuses on how the raw corpus was transformed into an ANT-style network of actors/actants. Texts were cleaned and tokenized (lowercasing, punctuation removal, and lemmatization), multiword expressions were preserved via n-gram extraction and named-entity recognition (NER), and term importance was assessed with TF–IDF via an empirical frequency cutoff (min document frequency = 3) to remove very rare terms while retaining semantically important items. The candidate terms were canonicalized via automated string matching and manual synonym grouping, and the top 300 candidate labels were independently checked. The terms were then classified into five actor types (human, organization/institution, place/toponym, artifact, practice/event) via rule-based cues (NER, POS patterns, suffixes) plus manual validation. Network nodes are these canonical, classified terms; edges derive from document-level co-occurrence with fractional counting and association-strength normalization applied before layout and clustering. The results therefore open with the ranked actor lists and descriptive statistics by type, followed by the constructed network visualizations and initial cluster labels that feed the ANT-informed interpretation.

5.2. Results of the Second Stage

This second stage builds on the findings of the previous stage, which focused on conceptualizing and visualizing the actant networks of Bey’s palace. This process involves analyzing the co-occurrence and total link strength among actants within the network data structure developed in the first stage. Various visualization techniques are employed, including network layout visualization, as illustrated in Figure 11, and actant density visualization, as shown in Figure 12.

The association strength algorithm analysis available in VOSviewer software [53] provides a clear visualization and an initial understanding of the key players that shape the network structure. This algorithm quantifies the relationships between actants by calculating a similarity score on the basis of their joint occurrences in relation to their individual frequencies. By doing so, it assesses the strength of the connection between two actants and visualizes those significant actants that exhibit strong relational ties.

This stage provides an initial response to our theoretical inquiry into the systematic application of ANT for mapping and analyzing the complexity of historic urban landscapes, with a specific focus on Bey’s palace, through the analysis of the characteristics and association quality of key actants and their agency in the dynamic network of Bey’s palace system. The ANT visualization depicts a complex network structure composed of both human and nonhuman actants, represented as nodes that signify actors, with edges illustrating the relationships among them. A notable aspect of this visualization is the distribution of weights, which illustrates the influence or agency of the actants within the network. This agency relates to the co-occurrence of actants within the structure and the overall strength of the connections that link them, serving as an attribute for each actant in the network.

The visualization graphs in Figure 11 highlight key actants through their centrality measures, which reflect total link strength and co-occurrence within the network. These central actants structure the system, and any disruption to them risks destabilizing the entire network. Within this configuration, Ahmed Bey emerges as the dominant focal actor, holding the highest betweenness, closeness, and degree centrality, thereby aligning diverse interests through translation [7]. Similarly, the Courtyard Garden, with the strongest total link strength, functions as an obligatory passage point (OPP), representing a condition all actors must navigate to achieve their goals.

Furthermore, the association strength algorithm has pinpointed essential actants crucial for maintaining network stability. As illustrated in Figure 12, which presents the actant density visualization, the density of each actant is determined by its co-occurrence with other actants in the network. Among these actants are places, palm trees, art, time, stories, walls, tiles, lights, windows, visitors, and columns. These actants play a significant role in constructing the network by mediating, engaging, and fostering deeper alliances among other actants within Bey’s palace network system.

Accordingly, this dynamic is explained by Latour [8] as a form of mediation that entails relationships or associations in which mediators exert influence over the actions of other actors, often in unpredictable ways. This phase corresponds to the interessement moment in the ANT translation process, where potential allies must be convinced to collaborate [65]. This stage provides a foundational analysis of the controversies surrounding the landscape of Bey’s palace, characterized by the significant influence of tangible elements. This analysis highlights the prominent features of affluence, luxury, and cultural significance that shape the landscape of Bey’s palace.

Accordingly, the results strongly support the research hypothesis by demonstrating the potential of association strength-based algorithmic analysis to identify pivotal roles of tangible and intangible actants, such as Ahmed Bey, the courtyard garden, and cultural narratives, in constructing and stabilizing the network.

Moreover, the findings reinforce the assertion made by ANT that systems of cultural heritage require diverse interconnections [22] by emphasizing the predominance of tangible actants alongside the importance of intangible elements such as time and tales. In addition, this outcome reinforces the idea that different types of actants are crucial for preserving the identity and functionality of historical urban systems. However, the identified controversies, particularly the risk of intangible cultural aspects being overshadowed by tangible elements, highlight the need for a revised perspective. Balancing these components could strengthen the sustainability of such networks, as reflected in Dobson’s [30] concept of sustainable change processes. In this context, Ahmed Bey emerges as a focal actor and the courtyard garden serves as an obligatory passage point (OPP); its actants including palm trees, fountains, water, cedar doors, and visitors, form strategic alignments that reinforce the site’s sociomaterial agency and support sustainable urban transformation.

5.3. Results of the Third Stage

5.3.1. First Step Results

In this stage, a comprehensive examination of the subsystems within Bey’s palace landscape network is conducted. This involves two complementary and synchronized algorithmic analyses: the LinLog layout algorithm, which visualizes clusters of actants on the basis of the strength of their connections, highlighting the successful enrollment of actants into roles within subsystems, and the Louvain modularity algorithm, which detects cohesive communities in the network that represent mobilized clusters of actants working together toward strong relational ties. These algorithmic analyses are executed via LinLog/Modularity analysis in VOSviewer 1.6.20 software [49].

Accordingly, as illustrated in Figure 13, through the application of the LinLog layout algorithm, the visual representation organizes the actants into seven subsystems or clusters according to their relationships. This visualization corresponds to the moment of enrollment in the ANT translation process, which involves persuading actors to accept specific roles within the network. This process requires aligning their interests and securing their engagement with the network’s objectives [17].

Additionally, the insights gained from applying the Louvain modularity algorithm, along with LinLog/modularity analysis via VOSviewer software, are taken into account. A computational analysis is conducted for each cluster, focusing on degree centrality measures that rely on two parameters: total link strength and total links for each cluster. This analysis is visually represented in a graph generated with the RAWGraphs tool, as illustrated in Figure 14. These clusters are classified on the basis of their shared characteristics as follows:

The wooden ornamentation and cultural expression subsystem; adaptive cultural heritage and artistic transformation subsystem; Ottoman and Islamic aesthetic synthesis subsystem; architectural mastery and cultural legacy subsystem; Arab-Andalusian atmosphere and cultural expression subsystem; cultural authenticity and artistic traditions subsystem; and aesthetic elegance and sensory design in the traditional architecture subsystem. These clusters or subsystems represent the actor-network concept in Bey’s palace landscape system and align with Cvetinovic et al.’s [18] definition of the concept of the actor network as interconnected assemblages that treat both humans and nonhumans symmetrically.

Additionally, this visualization illustrates that the Arab-Andalusian atmosphere and cultural expression subsystem and the architectural mastery and cultural legacy subsystem are the two most influential clusters, significantly impacting the whole network structure, according to their high ponderation among other clusters in the same network. They demonstrated that the interplay and connection between these clusters or subsystems of the Bey palace landscape system adopt a multiscale conservative approach to space-making inspiration.

In addition, as shown through the holistic overview in Figure 15, these two clusters play pivotal roles in preserving the local indigenous traditions of Constantine city at the local level while also embracing broader Arab and Mediterranean cultural values on a broader scale. The impact of these clusters is influenced by the agency of key actants within each cluster. Specifically, elements such as trees, sites, coolness, light, luxury, and courtyard gardens are essential components of the Arab-Andalusian atmosphere and cultural expression subsystem. These features encapsulate the essence of Arab and Mediterranean architectural inspiration.

Furthermore, tiles, columns, and windows are integral to the architectural mastery and cultural legacy subsystem, and the successful integration of these elements in Bey’s palace draws upon characteristics of local traditional architecture in Constantine city, including craftsmanship, painting, drawing, and the use of local materials. In addition, this result aligns with the perspective presented by Ernstson [25], which emphasizes the necessity of integrating various subsystems that connect the local historical context with extensive cultural and ecological networks in a historic landscape, Emstson defines this as the concept of interconnected conservation.

This result aligns with the mobilization moment of the ANT translation process, whereby all actors ensure that the collaborators adhere to specific control structures to maintain their agreed-upon actions [65].

From a sustainability perspective, clusters that combine civic actors, craft practices and place names suggest sociocultural networks that sustain local heritage value and constitute priorities for conservation and community support. High-centrality actors based on high total link strength (TLS) represent leverage points for policy or conservation protection, and engaging these actors can enhance adaptive capacity and continuity. Conversely, clusters with low internal cohesion or small TLSs may indicate fragile values and communities that would benefit from targeted interventions or participatory measures to bolster resilience.

5.3.2. The Second Step Results

This analysis step aimed to open the black box of Bey’s palace landscape system through computational analysis and graph visualization via the RAWGraphs tool [55]. This method enables the identification of key actants within the network of Bey’s palace landscape and their correlation with the principal actants involved in the ANT translation process. Our findings are based on a regression analysis that employs correlation graphs derived from two algorithmic task analyses produced by VOSviewer software: the eigenvector centrality algorithm and the co-occurrence algorithm. The analysis focused on three critical parameters: total links, total link strength, and the co-occurrence frequency of actants within the network structure as attributes.

Figure 16 shows the results of the eigenvector centrality algorithm analysis, through a regression analysis via the RAWGraphs tool, which is based on three key parameters: total links, total link strength, and the co-occurrence frequency of actants within the network. The two most optimized key actants identified in the entire network are the courtyard garden, recognized as the OPP, which has the highest degree of centrality, and Ahmed Bey, designated the focal actor, which has the second highest degree of centrality along with the highest closeness centrality to all other actors in the network. This is reflected in the graph, where lines connecting actants represent shared co-occurrence frequency measures within the network. Notably, Ahmed Bey shares the same co-occurrence frequency with both the courtyard garden and the other actants, indicating that he maintains equal proximity to all actants. This computational finding corroborates the earlier preliminary observations from the ANT visualization conducted in the second stage.

In addition, a separate regression analysis using the same previous tool is conducted on the basis of the results of the eigenvector centrality algorithm analysis, which focuses on each cluster within the network and utilizes the same previous parameters. Figure 17 illustrates the best actants in each cluster, identified by their highest degree of centrality. These actants serve as spokespersons during the mobilization moment of the ANT translation process [14]. Within each cluster, these spokespersons play a vital role by representing and advocating for various actors, which is crucial for the overall success of ANT [7].

In this case, several spokespersons are identified. These spokespersons shape different subsystems within Bey’s palace network structure. Specifically, Moorish architecture is a spokesperson in the wooden ornamentation and cultural expression subsystem; the courtyard garden is a spokesperson in the Arab-Andalusian atmosphere and cultural expression subsystem; Ahmed Bey is a spokesperson in the architectural mastery and cultural legacy subsystem; the wall is a spokesperson in the cultural authenticity and artistic traditions subsystem; Door is a spokesperson in the Ottoman and Islamic aesthetic synthesis subsystem; Decoration is a spokesperson in aesthetic elegance and sensory design in the traditional architecture subsystem; and the period is a spokesperson in the adaptive cultural heritage and artistic transformation subsystem. as shown in Figure 17.

This result validates Kim’s [31] dual strategies of stability and multiplicity within urban spaces. The role of Moorish architecture and the Arab-Andalusian atmosphere illustrates how spaces can remain stable in their historical essence while allowing for diverse and pluralistic interpretations that cater to a wide range of stakeholders. In an actor network, the spokesperson serves to unite various entities, referred to as intermediaries, to bolster their representations. Unlike human spokespersons, who can modify their goals [66] and are influenced by emotions [67], nonhuman spokespersons are more effective at maintaining network cohesion because of their unwavering presence. Within Bey’s palace landscape network, the identification of spokespersons, such as doors, walls, decorative elements, and Moorish architecture, further illustrates how these actants anchor the network. These nonhuman spokespersons exhibit consistency in preserving network stability that surpasses that of their human counterparts. This reliability is crucial in heritage landscape networks, where historical artifacts often symbolize enduring cultural values.

Furthermore, the results of the co-occurrence algorithm analysis conducted via VOSviewer software, along with the visual graph representation generated via the RAWGraphs tool through arc diagram network visualization [55], revealed various mediators. These mediators are vital in shaping and facilitating communication and progress within the landscape of Bey’s palace. As mentioned by Latour [8], a mediator is an entity that alters transforms, distorts, and affects the path of action during the Enrolement moment of the ANT translation process.

As illustrated in Figure 18, the identification of mediators is contingent upon their co-occurrence within the network, with frequency serving as the parameter across all clusters. The mediators include: the palm tree in the Ottoman and Islamic aesthetic synthesis subsystem; the fountain, water, and cedar wood doors in the aesthetic elegance and sensory design subsystem; the courtyard, site, space, and tree in the Arab-Andalusian atmosphere and cultural expression subsystem; the column, window, artist, and sculpture in the architectural mastery and cultural legacy subsystem; history transformation, art, and time in the adaptive cultural heritage and artistic transformation subsystem; tales, fresco legends, and painting in the cultural authenticity and artistic traditions subsystem; and the ceiling and masterpiece in the wooden ornamentation and cultural expression subsystem. These findings highlight the integration of both human and nonhuman agencies: nonhuman actants such as frescoes, legends, ceilings, and paintings operate as enduring cultural anchors that preserve continuity, whereas human actors, including artists, visitors, and stakeholders mobilize resources and narratives to adapt the system to evolving challenges. This duality reflects Allen’s [24] notion of synchronous heterogeneity in urban landscapes.

On the other hand, a fractional analysis in VOSviewer through a fractional counting algorithm [52] allows the identification of boundary objects that are artifacts, concepts, or terms that bridge social groups and enable coordination without full consensus. VOSviewer’s fractional counting assigns partial weights to co-occurrences (normalizing the influence of widely connected items) and, together with similarity scores and point-to-point links, highlights actors that stabilize the network and obscure internal complexity [17,53]. In Bey’s palace, the courtyard garden functions as a boundary object, underscoring the role of nonhuman actants in enrollment and network coherence. Figure 19 illustrates this boundary object.

5.4. The Integration and Discussion of These Results with Those of Previous Studies

The integration of the results of this research with those of previous studies from the literature review provides some insight as follows:

Dobson [30] and Ruming [22] underscore the importance of ANT in comprehending the interdependencies within urban systems, emphasizing the necessity for change and feedback loops to enhance urban resilience. Their emphasis on actor-networking and heterogeneous interconnections resonates with our study, which similarly highlights the dynamic relationships between human and nonhuman components. However, our research expands on these frameworks by incorporating computational algorithms to operationalize ANT, offering a more precise and scalable approach to visualizing actor networks in historic urban landscape HUL. In contrast to Dobson and Ruming, who focus primarily on conceptual networks, our study integrates practical tools for comprehensive network analysis.

Ernstson [25] examines the role of ANT in fostering resilient urban relationships, highlighting the need for a balance among social, ecological, and material agencies. His emphasis on framing and measuring value within networks aligns with our findings, particularly regarding the key actors in Bey’s Palace Complex as part of a broader network. However, a significant difference between his work and ours is that Ernstson does not incorporate computational tools, whereas our approach uses graph-theoretical algorithms to enhance translation and offer a dynamic representation of networks.

Allen [24] and Kim [31] employ ANT to examine urban landscapes, emphasizing human interactions with material spaces. Allen’s concept of “synchronous heterogeneity” and Kim’s exploration of stability and change within these environments are crucial for understanding the fluidity of actor-networks. Our approach builds on these studies by providing a computational framework that encompasses both the material and immaterial aspects of historic urban landscapes. This framework offers valuable insights into the intricate temporal and spatial dimensions that traditional methods often overlook.

Anne Tietjen [29] explored how ANT can create innovative urban landscapes by transforming agency components, emphasizing interventions that lead to new actor networks. While Tietjen focuses on intervention-driven innovation, our study is more concerned with systematically visualizing and mapping preexisting relationships in urban landscapes underscoring our methodological contribution.

Although actor–network theory (ANT) has long been used in social theory, architecture, and heritage studies, the specific computational operationalization of ANT’s translation process for historic urban landscape (HUL) analysis remains relatively limited. This apparent under exploration arises mainly from the field’s methodological history: ANT evolved within qualitative, ethnographic, and interpretive traditions that emphasize rich description over formalized modeling. Consequently, while numerous conceptual applications exist, relatively few studies have sought to translate ANT moments, including problematization, interessement, enrollment, mobilization, and stabilization into measurable computational processes [10].

A second explanation lies in the technical and infrastructural maturity of digital methods and network analysis tools. Only in the past decade have accessible and well-documented platforms such as VOSviewer [45,59] and RAWGraphs made reproducible controversy mapping and co-occurrence analysis practical for humanities and heritage researchers. Earlier studies that employed ANT for cultural heritage largely relied on manual coding or qualitative mapping [12,34], whereas current toolsets allow scalable analysis of textual and archival corpora, enabling the hybrid ANT and graph-theoretic workflow demonstrated in this study.

Practical barriers have also contributed to the scarcity of such work. Implementing a computational ANT framework requires interdisciplinary expertise that bridges STS, natural-language processing, and network science, as well as access to sufficiently rich textual/archival data and methodological justification for parameter settings such as the counting method and normalization [64]. These factors help explain why previous research often stopped short of full algorithmic implementation.

Within this context, the present study’s contribution is primarily methodological. By offering a transparent, step-by-step framework including data collection, preprocessing, term extraction, network construction, clustering/centrality analysis, and interpretation together with explicit software parameters (see Table 1A–C), we respond to the methodological gap identified above. The approach builds upon recent interdisciplinary efforts that link ANT with digital methods and urban data studies [35,41,44,59] and demonstrates how such integration can produce interpretable, policy-relevant insights for sustainable heritage management. Future comparative and mixed-methods studies are encouraged to test and extend this framework across other HUL contexts.

To summarize the findings of this research, a holistic circular graph of ANT was created via the RAWGraphs tool. This graph enables us to decode the landscape system of Bey’s palace and emphasizes the influence of key actors who shape its sociotechnical network, as illustrated in Figure 20.

5.5. Research Generalizability and Limitations

While this research offers a methodological contribution, several limitations must be acknowledged. First, the research is based on a single-case example (Bey’s Palace), which constrains direct empirical generalization across all historic urban landscapes (HULs). Second, the corpus used for analysis is composed of available textual and archival sources in specific languages and media formats, which may bias the network representation toward well-documented actors and discourses. Third, network outcomes depend in part on analytic choice preprocessing steps, counting methods, normalization, clustering resolution and layout parameters and different choices can yield variations in cluster boundaries and centrality rankings. Fourth, this study relies on retrospective textual sources and does not include live sensor or participatory real-time data, which limits temporal insight into mobilization dynamics.

However, none of these limitations invalidate the paper’s principal contribution. The core aim here is methodological: to operationalize ANT’s translation process as a transparent and replicable framework (data collection, preprocessing, term extraction, network construction, clustering/centrality analysis, interpretation). The value of such a framework lies in making choices explicit, repeatable and testable. To that end, we took several steps to increase robustness and transparency: (a) we documented preprocessing choices and term-thresholds; (b) we used fractional counting and association-strength normalization to reduce bias from highly prolific sources; (c) we reported parameter settings for clustering and layout and provided sensitivity notes where appropriate; and (d) we triangulated quantitative outputs with targeted archival observations and the palace’s empirical description to support interpretation. These actions ensure that network measures are treated as interpretive indicators rather than definitive causal proofs.

In summary, the limitations relate mainly to the scope and available data rather than to flaws in the conceptual or computational design. Because the framework’s structure and implementation choices are explicitly reported and defensible, the study remains a valid and useful methodological contribution. It provides a practical platform for future comparative and mixed-methods research that will progressively address the limitations noted above.

5.6. Research Study Implications

The findings reported in this study have implications at several interrelated levels: theoretical, methodological, and practical/policy oriented. Below, we articulate these implications and indicate avenues for further development and empirical testing.

Theoretical implications and contribution. This research contributes to debates about agency, mediation and spatial agency in historic urban landscapes by demonstrating a concrete way to translate actor–network theory’s conceptual vocabulary into measurable, interpretable indicators. By operationalizing ANT’s translation moments (problematization, interessement, enrollment, mobilization, and stabilization) through a sequence of natural language processing, co-occurrence normalization, community detection and centrality measures, this study makes two theoretical moves. First, it shows how the distributed agency of heterogeneous actants can be traced empirically in textual and archival corpora and represented as network properties (for example, cluster composition and stability, centrality and bridging scores), thereby making ANT’s claims about relational power and mediation empirically tractable. Second, by explicitly identifying obligatory passage points and boundary objects through a combination of centrality/bridge metrics and fractional counting, the paper clarifies the mechanisms by which material and discursive items operate as mediators in both stabilizing and transforming urban networks. Together these moves bridge a gap between interpretive ANT scholarship and computational network science, offering a theoretically informed, empirically grounded account of how sociotechnical relations in HULs can be mapped and interrogated.

Methodological implications. The principal methodological contribution is a transparent, transferable framework that sets out clear choices at each step from corpus construction and preprocessing to normalization, clustering and interpretation so that both scholars and practitioners can replicate and adapt the procedure in other HULs. Explicit reporting of choices (for example, the counting method, association–strength normalization, clustering resolution and minimum term frequency) is central to ensuring interpretive validity and comparability. The approach also highlights two critical methodological recommendations for future work: (1) combine quantitative network outputs with small-scale qualitative validation (interviews, targeted archival checks) to ground network interpretation in context; and (2) report robustness checks (sensitivity to normalization and clustering parameters) so that claims about OPPs, mediators and spokespeople are demonstrably stable. By lowering the technical and epistemological barriers to combining ANT sensibilities with digital-methods, this framework contributes to a more cumulative and reproducible research program in HUL studies.

Practical and policy implications for heritage managers and urban practitioners. This framework furnishes sustainability-relevant diagnostics that can be integrated into monitoring frameworks for SDG11 (sustainable cities and communities). The proposed framework provides heritage managers with diagnostic outputs that can be integrated into decision frameworks: (a) use centrality/TLS measures to prioritize resource allocation and intervention, (b) use cluster stability analyses to identify vulnerable social/heritage ensembles requiring community support, and (c) use identified OPPs to design governance interventions or policy levers. These operational linkages make the method suitable for incorporation into sustainability monitoring frameworks and contribute to evidence-based stewardship of historic urban landscapes.

Furthermore, centrality and total link strength identify candidate focal actors and potential OPPs that may warrant conservation or mediation, and cluster composition and stability measures suggest which ensemble features and narratives are most central to sustaining community value and which are vulnerable. Presenting these outputs in accessible visual formats (cluster maps, actor-term dashboards) can support participatory decision-making and help translate complex controversy maps into concrete interventions for example, targeted maintenance of boundary objects identified as essential connectors, or mediated stakeholder dialogues around identified spokespersons and OPPs.

5.7. Recommendations for Future Research

Future research should move beyond demonstration and increase external validity, and we recommend and plan the following research steps. First, the framework is applied to a comparative set of 3–5 HUL cases chosen for contrasting governance, scale and data availability to test the method’s portability. Second, integrate small-scale qualitative validation (semi structured interviews, archival cross-checks) with network outputs to ground interpretive claims. Third, temporal data, either longitudinal textual corpora or sensor-derived signals, are essential for studying mobilization dynamics and the persistence of obligatory passage points (OPPs) over time. Fourth, we report and publish robustness checks (sensitivity of clusters and centrality to normalization and resolution parameters) as a standard practice to allow readers to evaluate stability. Finally, we encourage open sharing of preprocessing scripts and cleaned corpora where ethics and privacy permit, to facilitate replication.

6. Conclusions

This study has demonstrated the vital role of ANT and computational analysis tasks based on graph-theoretical algorithms from network analysis tools in revealing the complexities of historic urban landscapes. Focusing on Bey’s palace complex in Constantine, Algeria, a site imbued with rich cultural and historical significance, this research has provided a methodological innovation that bridges the gap between theoretical frameworks and practical applications for historic urban landscape analysis.

The findings highlight the importance of the diverse components, both human and nonhuman, that make up historic urban landscapes. These elements encompass tangible aspects, such as Moorish architecture, decorative features, and spatial configurations, as well as intangible factors, such as historical memory, cultural values, tales, and symbolic figures such as Ahmed Bey. These interconnected actors form dynamic networks that collectively shape the urban landscape and help sustain its continuity over time.

By using the mapping controversies technique in conjunction with network analysis tools such as VOSviewer software and RAWGraphs, four distinct algorithmic tasks were performed: the natural language processing algorithm (NLPA), the Louvain modularity algorithm, and two types of force-directed algorithms, which are the association strength algorithm and the LinLog layout algorithm. Moreover, computational analyses were conducted via the eigenvector centrality algorithm and the co-occurrence algorithm.

This application of ANT has been successfully transformed into a practical framework that enhances the visualization and analysis of historical urban landscape HUL networks. This approach effectively illuminates the complexities of the urban landscape, allowing us to identify key actors and key clusters and their interactions to understand their roles in sustaining the stability, adaptability, and cultural vitality of Bey’s palace.

The algorithmic analysis tasks revealed not only the strength of dominant actors/actants, such as Ahmed Bey, courtyard gardens, and symbolic architectural elements but also the critical role of most key clusters or actor networks, such as the Arab-Andalusian atmosphere and cultural expression subsystem, and the architectural mastery and cultural legacy subsystem. These key actors/actants and the actor networks significantly impact and connect the whole network structure and enhance the spatial agency within Bey’s palace system.

Importantly, this research validates the hypothesis that integrating ANT’s translation process through the mapping controversies technique with graph-theoretical algorithms offers a superior framework for analyzing the intricate relationships within historic urban landscapes. The results show how these procedural tasks could enhance the ability of architects, urban planners, and heritage professionals to understand, visualize, and engage with these spaces, fostering innovative and culturally sensitive approaches to their preservation and regeneration. The study also highlights the duality of historic urban landscapes as spaces of both stability and transformation. By leveraging the interconnectedness of diverse actors, urban landscapes can maintain their historical continuity while simultaneously adapting to contemporary needs. This balance is crucial for ensuring that historic centers remain dynamic cultural assets that resonate with their past and present contexts.

This research makes a substantial contribution to the fields of urban studies, architectural theory, and heritage management by offering a solid theoretical and practical framework for architects and urban planners to analyze and engage with historic urban landscapes. This study provides actionable recommendations for the management of these landscapes. By highlighting the importance of integrating diverse actors and networks, a holistic approach to urban development that values diversity, fosters creativity, and preserves the distinctive identities of historic centers is advocated. Beyond theoretical advances, this work delivers a practical toolkit that supports sustainability objectives for historic urban landscapes by enabling measurable prioritization, monitoring and governance of cultural, social and infrastructural assets.

This research has certain limitations, particularly its focus on a single case study within a specific context. As a result, the applicability of the findings to other urban environments remains uncertain. Future studies should expand upon this methodological framework, investigating its relevance in different contexts and incorporating additional advanced computational tools and digital mapping techniques: