An estimated 10-20% of the world's population, up to 1.2 billion people, belong to ethnic and cultural minorities, many of whom face discrimination, marginalization, and systemic exclusion. These realities underscore the urgent need for inclusive, culturally sensitive systems that safeguard and accurately reflect minority identities (Goodman et al., 2023). However, despite decades of progress in information science, many existing knowledge systems fail to capture the complexity, fluidity, and richness of these identities. The central question thus arises: how can we ethically and accurately represent ethnographic diversity in the digital systems upon which researchers, institutions, and policymakers increasingly rely?

Knowledge Organization Systems (KOS), including taxonomies, classification schemes, thesauri, and ontologies, have long provided the backbone for structuring and retrieving information (Feng et al., 2024; Moraitou et al., 2022). Yet these systems, often shaped by Western epistemological frameworks, embed structural biases that marginalize non-Western and Indigenous knowledge (Doyle, 2013; Rubim Silva & Dal'Evedove, 2025). As a result, traditional KOS tend to represent knowledge from dominant cultural standpoints, rendering minority perspectives either invisible or misrepresented (Wright & Saelua, 2023; Anania & Stiglitz, 2023).

In contrast, Knowledge Graphs (KGs) offer semantically rich and flexible structures for modeling complex relationships between entities. As graph-based systems, KGs are increasingly recognized for their potential to support pluralistic and interdisciplinary knowledge ecosystems, particularly when integrated with culturally grounded ontological frameworks (Dally et al., 2024; Nguyen et al., 2023). Moreover, language is a central dimension of both KOS and KGs. Recognizing and integrating minority languages not only supports cultural preservation but also promotes equitable access to services, data, and epistemic justice (Couldry & Mejias, 2023).

Nonetheless, mainstream classification systems such as the Library of Congress Subject Headings (LCSH) and the Dewey Decimal Classification (DDC) continue to draw criticism for their Eurocentric hierarchies. In response, alternative and community-led models like Lau ā Lau ka ʻIke and the Infomediary of Taiwanese Indigenous Peoples have emerged to reflect Indigenous epistemologies (Sung & Chi, 2021; Wright & Saelua, 2023). While these initiatives exemplify valuable resistance to epistemic hegemony, they remain peripheral in the global knowledge infrastructure (Huang & Zaslavsky, 2024). Moreover, challenges persist regarding data sovereignty, co-design ethics, and the risk of reproducing power asymmetries through externally imposed ontologies (Siegers et al., 2023; Hou et al., 2022).

Recent efforts to document intangible heritage, such as languages, oral histories, and cultural artifacts, have demonstrated the value of linked open data (LOD) and KG technologies (Cui et al., 2024; Du et al., 2025; Fan, 2023). However, many of these systems still fall short in representing hybrid or evolving identities, especially when modeled through rigid taxonomies. For example, ontologies developed for health or archival data, while improving formal data accuracy, often struggle to reconcile contextual, relational, and narrative-based forms of knowledge (Nguyen et al., 2023; Mateos, 2007; Woods, 2017).

In the domain of knowledge management, the role of tacit knowledge, oral traditions, and community archiving remains central, particularly among minority groups adapting to globalization and generational change (Sewdass, 2014; Bonilla et al., 2025). Nevertheless, structural inequities, microaggressions, and digital divides continue to impede effective knowledge sharing and cultural visibility (Davis, 2015; Khurram & Giangiulio, 2020; Li et al., 2023).

Therefore, Emerging Linked Open Data (LOD) and knowledge graph applications offer promising solutions. Projects in Laos, Vietnam, and the Greater Mekong Subregion have enabled multilingual, standardized access to ethnic knowledge and folktales (Chansanam et al., 2024; Ngootip et al., 2023; Ngoc & Chansanam, 2023). Broader efforts also document applications in preserving craftsmanship, biodiversity, and ethnographic knowledge (Bai & Hou, 2023; Chansanam et al., 2022, 2020; Cui et al., 2024; Lu et al., 2023). Still lacking, however, is a cohesive, interdisciplinary synthesis that critically evaluates the equity and sustainability of these digital innovations (Jaroenruen et al., 2024; Khoo et al., 2024).

To address this gap, this study engages with both decolonial information science and intercultural communication theories. Drawing on scholars such as Mignolo (2011) and Green (1999), it recognizes the need for research methodologies that legitimize Indigenous knowledge structures and allow for community-defined ontologies. Additionally, perspectives from Gudykunst (2003, 2004) and Kim (2015) shape the system’s emphasis on relational modeling, multilingual design, and cultural fluidity, ensuring that identity representation is negotiated rather than imposed.

Despite advances in graph technologies, traditional KOS continue to rely on siloed, hierarchical taxonomies that restrict the representation of complex ethnographic relationships (Smith, 2021). This structural rigidity not only limits interoperability across databases but also impedes discovery, learning, and cross-disciplinary innovation. Consequently, there is a pressing need for systems that can semantically integrate geo-temporal, linguistic, cultural, and relational data in a unified yet flexible framework.

This study responds to these challenges by designing, constructing, and evaluating a knowledge graph that models the ethnographic diversity of 375 ethnic groups across six countries in the Greater Mekong Subregion (GMS). The graph incorporates multilingual identifiers, migration histories, cultural practices, and linguistic lineages, offering a nuanced and scalable model for inclusive information representation. Furthermore, the project includes a custom-built web application for real-time visualization, interactive exploration, and expert validation. Therefore, this research addressed the following research question, how can a semantically enriched, ethically grounded knowledge graph framework improve the representation, accessibility, and interpretability of complex ethnographic data across the Greater Mekong Subregion?

However, this study makes four key contributions. First, it introduces a cross-border, region-wide KG on ethnographic diversity, integrating community-sourced and validated data. Second, it advances methodological integration by combining KOS, KG, and knowledge management approaches to model identity, migration, and cultural relationships. Third, it presents an open-access web platform supported by robust expert validation and usability metrics. Finally, it establishes a reusable framework for inclusive, scalable, and ethically oriented digital heritage systems that can be adapted for other regions and disciplines. Moreover, this research not only contributes to digital ethnography and semantic technologies but also advocates for epistemic justice by challenging dominant paradigms in knowledge organization and representation.

This research focuses on developing and implementing a Knowledge Graph (KG) web application designed to model the cultural and ethnic diversity of the Greater Mekong Subregion (GMS). The study emphasizes the transformation of a structured dataset into a semantic knowledge graph that captures the complex relationships among 375 ethnic groups across six countries, Thailand, Laos, Myanmar, Cambodia, Vietnam, and China, highlighting linguistic, cultural, geographic, and historical attributes. The construction process for the GMS Ethnic Groups Knowledge Graph is illustrated in Figure 1.

The results of this study present the structure, content, and analytical capabilities of the Greater Mekong Subregion (GMS) Ethnic Groups Knowledge Graph (KG) and its associated web application. Through a series of visualizations, database queries, and interface demonstrations, the findings illustrate how ethnographic data for 375 ethnic groups across six countries have been semantically modeled and interactively explored. These results highlight the knowledge graph’s ability to represent complex interrelationships among cultural, linguistic, religious, and geographic dimensions. Detailed visual outputs, such as ego-network views for specific ethnic groups and geospatial mappings of ethnic distributions, demonstrate the system’s capacity to capture multidimensional identities and support domain-specific research. Additionally, technical and user-centered performance evaluations validate the knowledge graph’s practical utility and superiority over traditional database approaches in representing ethnographic complexity.

The general characteristics of GMS Ethnic groups knowledge graph

This study investigated the potential of semantic knowledge graphs (KGs), ontologies, and Knowledge Organization Systems (KOS) to ethically represent ethnographic data across the Greater Mekong Subregion (GMS). The GMS Ethnic Groups Knowledge Graph successfully integrated data on 375 ethnic groups across six countries, modeling intricate linguistic, cultural, religious, and migratory relationships. Compared to conventional relational databases, the KG demonstrated significant improvements in usability, interpretability, and analytical capabilities, as evidenced by over a 25% increase in task performance metrics during expert evaluations.

These results substantiate growing calls for more inclusive and context-aware knowledge infrastructures (Doyle, 2013; Rubim Silva & Dal'Evedove, 2025; Wright & Saelua, 2023). Unlike traditional classification systems such as the Library of Congress Subject Headings (LCSH) or the Dewey Decimal Classification (DDC), the KG presented in this study prioritizes marginalized perspectives, advancing epistemic justice through culturally responsive and community-informed design. While the project builds upon prior initiatives, including Lau ā Lau ka ʻIke (Wright & Saelua, 2023) and Linked Open Data (LOD) projects in Laos and Vietnam (Chansanam et al., 2024; Ngootip et al., 2023; Ngoc & Chansanam, 2023), it extends these contributions by offering a scalable, multi-country implementation that features interactive filtering, temporal migration tracking, and domain-specific validation.

Scholars have long argued that established knowledge structures often perpetuate the marginalization of non-Western epistemologies (Doyle, 2013; Tran, 2017). In response, Indigenous information theorists have advanced alternative frameworks such as community-controlled vocabularies, participatory classification systems, and relational metadata structures (Dlamini, 2024; Smith, 2021; McDonough, 2013; Tuhiwai Smith, 2012). These approaches emphasize the value of local knowledge authority, oral traditions, and cultural sovereignty in the development of knowledge infrastructures.

Aligned with this critical discourse, decolonial information science advocates for dismantling the hierarchies embedded in metadata schemas, authority control mechanisms, and archival practices (Ntshoe, 2020; Sugimoto & Wijesundara, 2024). The present study contributes to this movement by employing a semantic architecture that enables fluid, multilingual, and self-determined identity modeling. This design facilitates the co-existence of plural ontologies and supports the layered representation of ethnicity, language, and cultural practices.

Despite the innovations presented, several limitations must be acknowledged. Most notably, the data coverage for historical migration patterns and syncretic religious practices remains incomplete. This limitation reflects the intrinsic difficulties in sourcing and verifying such information, as well as the gaps inherent in available historical records. Furthermore, while the platform’s visualization and interaction features are robust, the reliance on manually curated datasets poses challenges to scalability. Future enhancements should therefore consider the integration of natural language processing techniques and automated entity recognition to support expansion. Moreover, the lower completeness scores observed in specific categories underscore the necessity of deepened collaboration with local ethnographers and cultural experts.

The manual nature of the curation process, though critical for ensuring cultural sensitivity, introduces an element of subjectivity, particularly in classifying dynamic cultural phenomena such as religious practices and festivals. These representational challenges are symptomatic of broader epistemological tensions. As Doyle (2013) and Tran (2017) emphasize, structured classification systems may inadvertently essentialize fluid and hybrid identities. While the knowledge graph developed in this study is designed for flexibility and inclusivity, it still simplifies lived experiences to accommodate computational representation. Compared to initiatives such as Lau ā Lau ka ʻIke (Wright & Saelua, 2023) and Taiwan’s Indigenous Infomediary project (Sung & Chi, 2021), this research expands the geographic scope and technical scale while maintaining a strong commitment to participatory, context-sensitive design. It seeks to balance the authenticity of community representation with the demands of interoperability and digital infrastructure.

Nonetheless, an unresolved tension persists between the structured nature of knowledge graphs and the inherently dynamic character of cultural identity. Ethnic, linguistic, and religious affiliations often shift over time and resist fixed categorization. Although this study incorporates multilingual labels, temporal filters, and relational structures to mitigate essentialism, no digital system can fully encapsulate contested or evolving identities. Therefore, continued community engagement remains critical for ensuring the long-term integrity and ethical validity of the system.

This study offers three principal implications. First, it presents a replicable framework for the digitization and semantic modeling of ethnographic knowledge in other regions, thereby contributing to global efforts in inclusive knowledge representation. Second, it demonstrates that integrating KOS principles with knowledge graph architectures can simultaneously support computational efficiency and cultural nuance. Third, the developed application serves as a valuable tool for interdisciplinary education and research, with practical applications in digital humanities, anthropology, linguistics, and data science.

Ultimately, the findings reaffirm the study’s central proposition: that mainstream knowledge systems often fail to capture the complexity of minority identities, necessitating the development of ethically grounded alternatives (Doyle, 2013). By drawing on interdisciplinary methods and community-sourced data, this project challenges dominant paradigms and contributes to ongoing debates on equity, justice, and inclusion in digital knowledge infrastructures (Khoo et al., 2024; Siegers et al., 2023).

The GMS Ethnic Groups Knowledge Graph illustrates how semantically rich, culturally inclusive systems can advance ethical knowledge representation. It promotes equitable access, deeper cultural understanding, and sustainable heritage documentation. Looking ahead, future work should prioritize sustained collaboration with local communities to ensure that marginalized voices continue to shape and define the narratives embedded within digital infrastructures.

Looking ahead, future work should prioritize sustained collaboration with local communities to ensure that marginalized voices continue to shape and define the narratives embedded within digital infrastructures. Additionally, I recommend elaborating more on how end-users, such as policymakers, educators, or ethnic communities, might directly engage with the platform in real-world contexts. Clarifying use scenarios can help translate technical outcomes into meaningful social impact and guide future system development and adoption.

This study developed and evaluated a semantic knowledge graph (KG) and accompanying web application that represents 375 ethnic groups across the Greater Mekong Subregion. By integrating culturally grounded data with semantic web technologies, the project demonstrated that knowledge graphs can model complex cultural, linguistic, religious, and geographic relationships with greater flexibility and nuance than traditional databases. Expert validation and performance metrics confirmed substantial improvements in usability, query precision, and knowledge discovery. A core contribution of this work lies in its interdisciplinary synthesis of Knowledge Organization Systems (KOS), Knowledge Management (KM), and Digital Humanities. This integration responds to growing calls for more equitable, inclusive, and context-sensitive digital infrastructures that center minority and Indigenous knowledge systems. Unlike earlier initiatives limited in geographic or technical scope, this region-wide implementation supports multilingual navigation, temporal analysis, and participatory validation.

However, some limitations remain. The manual nature of data curation poses scalability challenges, while the classification of evolving or hybrid cultural practices introduces representational ambiguities. Additionally, the dataset’s coverage of historical migration and syncretic religious practices is incomplete and requires further enrichment. Looking ahead, future developments should explore the integration of natural language processing for scalable entity extraction, incorporate user feedback for dynamic ontology refinement, and expand both linguistic and geographic coverage. Most importantly, sustained collaboration with local communities is essential to ensure authenticity, cultural appropriateness, and epistemic justice.mUltimately, this framework offers a replicable foundation for building ethical, pluralistic knowledge infrastructures in the domain of digital cultural heritage. It has the potential to inform not only academic research but also educational tools, policymaking, and community-driven documentation initiatives. This study contributes to the field of intercultural communication by offering a semantically rich, culturally grounded platform that enhances mutual understanding, models identity fluidity, and supports equitable dialogue across diverse cultural groups.

Acknowledgement Statement: The authors would like to thank to all participants and the reviewers for providing comments in helping this manuscript to completion.

Conflicts of interest: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Authors' contribution statements: Wirapong Chansanam, Lan Thi Nguyen, Chunqiu Li, and Christopher Khoo Soo Guan performed the experiment. Wirapong Chansanam and Lan Thi Nguyen wrote the manuscript with support from Chunqiu Li and Christopher Khoo Soo Guan .Wirapong Chansanam and Lan Thi Nguyen fabricated the GMS ethnic groups' dataset, and Wirapong Chansanam supervised the project.

Funding statements: This research was funded by the Faculty of Humanities and Social Sciences, Khon Kaen University, grant number HUSO-2568.

Data availability statement: Data is available at request. Please contact the corresponding author for any additional information on data access or usage.

Disclaimer: The views and opinions expressed in this article are those of the author(s) and contributor(s) and do not necessarily reflect JICC's or editors' official policy or position. All liability for harm done to individuals or property as a result of any ideas, methods, instructions, or products mentioned in the content is expressly disclaimed.