Music, Heritage, and Policy in the Age of AI

Music, Heritage, and Policy in the Age of AI Building the Diversity Pillar of the European Music Observatory with Federated Knowledge Graphs Table of contents Abstract....................................... 1 Introduction........................................ 3 Findings of the Feasibility Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Policy Context and the European Data Governance Agends . . . . . . . . . . . . . 7 Methodological Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Diversity Goals Without Exposure of Sensitive Data . . . . . . . . . . . . . . . . . 18 Application Layers and System Architecture . . . . . . . . . . . . . . . . . . . . . . 20 Case Study: Computing Local Content Indicators from Official Charts . . . . . . . 22 Datasourcesandscope .............................. 22 Rationale for the use of official charts as a neutral reference corpus . . . . . . 23 Microdata layer: recordings and works . . . . . . . . . . . . . . . . . . . . . . 24 Internal evaluation and indicator logic . . . . . . . . . . . . . . . . . . . . . . 24 Aggregation and observatory-level representation . . . . . . . . . . . . . . . . 25 Interpretation limits and policy relevance . . . . . . . . . . . . . . . . . . . . 25 DiscussionandOutlook ................................. 25 References......................................... 27 Abstract Across Europe and other advanced music markets, cultural policy instruments such as local content quotas, diversity targets, and equality pledges are widely used to sustain domestic cultural ecosystems. Despite their long history, these instruments have repeatedly struggled at the point of implementation. Broadcasters, festival organisers, and other cultural intermediaries are expected to meet quantitative targets, yet they often lack reliable, auditable, and legally safe tools to assess compliance in practice—particularly under the constraints imposed by data protection law. This paper presents the Slovak Comprehensive Music Database (SKCMDb) as a concrete response to this challenge. Developed as part of the Open Music Observatory, SKCMDb operationalises diversityand locality-related policy criteria through a federated, governance1 by-design data infrastructure. Rather than exposing personal attributes of composers or performers, the system separates internal evaluation from external disclosure, expressing policy-relevant outcomes as properties of cultural artefacts. This approach enables the computation of eligibility and diversity indicators while respecting the General Data Protection Regulation, non-discrimination principles, and institutional competences. The paper situates SKCMDb within a broader ecosystem of interoperable modules, including the Hungarian Music Database and community-curated Livonian and Karelian music databases, illustrating how subsidiarity can be implemented across national, community, and supra-national levels. Methodologically, the work combines linked-data modelling, modular ontologies, and provenance-aware workflows with lightweight publication strategies. An empirical case study based on 156 weeks of official Slovak music charts demonstrates how local content indicators can be computed reproducibly and transparently using authoritative reference data, without monitoring individual broadcasters. The results illustrate the feasibility of diversity-aware cultural policy infrastructures that support voluntary compliance, editorial decision-making, and comparative analysis. The paper concludes by outlining implications for European data governance, cultural policy, and the future development of the Open Music Observatory as a prototype European Music Observatory. ĹNote This is an early draft of the D2.1 deliverable of the Open Music Europe consortium. It has not been approved, amended, consulted with many contributors, and therefore it is not intended for use or citation. We make this version available in compliance with the Open Policy Analysis guidelines. The latest working copy is available here. Transparency note: Importantly, this document does not present a purely conceptual or speculative policy design. Substantial parts of the analysis synthesis lessons from policy-embedded implementation activities carried out using Open Policy Analysis methods1, including openly documented data, code, workflows, and methodological decisions. These implementations were conducted in cooperation with public authorities and sectoral institutions, allowing the Green Paper to reflect not only policy aspirations but also the practical constraints and opportunities observed in real-world deployment. Following the principles of Open Policy Analysis, all related deliverables and technical documentation are publicly accessible to foster broad engagement and ensure a clear audit trail. Supporting documents for each chapter of this Green Paper are referenced in similar boxes. The current version (and future White Paper drafts) is available at https://zenodo.org/records/17075796. Standardised folders, figures, and bibliographies are available at https://github.com/dataobservatory-eu/open-music-data-whitepaper. Please note that this document puts the Open Music Observatory, a prototype of a modern European Music Observatory developed by the OpenMusE consortium, which is being currently populated with economy, diversity, society, innovation data and has already three federated modules, can reviewed in the technical documentation (see versioned Zenodo DOIs), our viewn on the temporary landing page. 2 Funding acknowledgement: This project has received funding from the European Union’s Horizon Europe programme under Grant Agreement No. 101095295. The views expressed are those of the authors only and do not necessarily reflect those of the European Commission or its agencies2. Citation note: When citing this document, please use the latest versioned DOI available on Zenodo (current version: 10.5281/zenodo.17922390), and include the date of access if referring to material hosted on our GitHub repository.3This is an early 0.1.0 version.) Introduction Across Europe and other advanced music markets, cultural policy instruments such as local content quotas, diversity targets, and equality pledges have long played an important role in sustaining domestic cultural ecosystems. These instruments reflect a structural asymmetry between global cultural markets and national or regional creative production. While Englishlanguage popular music circulates globally with few barriers, music produced in smaller linguistic and cultural markets—such as Slovak music—is consumed predominantly within national borders or among relatively small diaspora communities. For this reason, more than eighty countries worldwide apply some form of preferential treatment to local music and audiovisual production, most commonly through minimum usage shares for public broadcasters and, in many cases, commercial, regional, or community radio stations as well. Despite their long history, these policy instruments have repeatedly struggled at the point of implementation. Broadcasters, festival organisers, and other cultural intermediaries are formally expected to meet quantitative targets, yet they often lack reliable, auditable, and ethically acceptable tools to assess compliance in practice. In the absence of such tools, 3The Guidelines for Open Policy Analysis form an actionable and practical set of directives that the OpenMusE consoritum was mandated to use under the Grant Agreement. It can be seen as good implementation framework for Evidence-based policy making in the European Commission and The practice of reproducible research [BITSS (2019); Open Music Europe (2023); (J 2015; Kitzes, Turek, and Deniz 2018). 3This document has been prepared by Open Music Europe (OpenMusE) project partners as an account of work carried out within the framework of this contract. Any dissemination of results must indicate that it reflects only the author’s view and that the Commission Agency is not responsible for any use that may be made of the information it contains. Neither Project Coordinator, nor any signatory party of Open Music Europe (OpenMusE) Project Consortium Agreement, nor any person acting on behalf of any of them: (a) makes any warranty or representation whatsoever, express or implied, (i) with respect to the use of any information, apparatus, method, process, or similar item disclosed in this document, including merchantability and fitness for a particular purpose, or (ii) that such use does not infringe on or interfere with privately owned rights, including any party’s intellectual property, or (iii) that this document is suitable to any particular user’s circumstance; or (b) assumes responsibility for any damages or other liability whatsoever (including any consequential damages, even if advised of the possibility) resulting from your selection or use of this document or any information, apparatus, method, process, or similar item disclosed herein. 3Always use the latest versioned DOI when citing this document, available via Zenodo. If you rely on supporting material hosted in the GitHub repository, please add the date of access in your reference. 3 implementation has frequently relied on informal editorial heuristics, institutional memory, or conservative programming strategies that privilege already well-known repertoire. These practices, while understandable, tend to undermine the intended goals of cultural diversity, renewal, and inclusion. In the European context, the introduction of the General Data Protection Regulation (GDPR) has further sharpened this implementation challenge. Many contemporary diversity-oriented policy instruments—including quotas aimed at supporting women artists, young creators, or local and European repertoire—implicitly assume that editors or curators can access and process personal attributes of composers and performers, such as age, gender identity, nationality, ethnicity, or place of residence. Initiatives such as the KeyChange pledge, which seeks to improve gender balance on festival stages and in concert programming, similarly presuppose access to identity-related information. Yet these attributes belong precisely to those categories of personal data that are subject to heightened protection under GDPR. As a result, the practical monitoring and enforcement of such targets risk colliding with fundamental principles of European data protection law. Crucially, the monitoring problem predates GDPR. Even before the regulation entered into force, the rapid expansion of global music distribution platforms had already rendered traditional editorial knowledge insufficient. In the 1990s, a highly experienced radio editor or festival curator in a smaller European country might plausibly have maintained working familiarity with a few tens of thousands of recordings circulating through physical distribution channels. Today, global digital platforms provide access to well over one hundred million recordings. No individual editor can meaningfully survey such a repertoire, yet cultural policy objectives increasingly apply across this vastly expanded universe of available content. This scale effect fundamentally alters the nature of categorisation, assessment, and compliance. These developments expose a structural mismatch between cultural policy ambitions and the infrastructures available to support them. The problem is not primarily one of unwillingness or resistance among cultural intermediaries, but of missing data coordination mechanisms capable of translating policy criteria into operational, computable form. At the same time, any viable solution must respect a core principle of European governance: subsidiarity. Cultural production, heritage, and identity are inherently situated, and meaningful data infrastructures must therefore operate at multiple levels—local, national, and supra-national—without collapsing these distinctions into a single, centralised system. The work presented in this paper emerges from this insight. It introduces the Slovak Comprehensive Music Database (SKCMDb) as a concrete national-level response to the challenge of operationalising diversity-aware cultural policy under conditions of scale and data protection. SKCMDb was developed alongside comparable initiatives, notably the Hungarian Music Database (HUNDB), which addresses similar policy questions within a different institutional ecology. While both systems rely on linked-data principles and federated governance, their emphases differ. The Slovak implementation is more closely aligned with library and bibliographic traditions, reflecting the central role of national libraries and collective management organisations. The Hungarian implementation, by contrast, places greater 4 weight on archival collections and heritage-oriented sources, mirroring the stronger role of music archives and ethnomusicological repositories. Beyond the national level, the same subsidiarity-driven logic extends both downward and upward. Within the Finno-Ugric Data Sharing Space, smaller, community-curated databases— such as the Livonian and Karelian Music Databases—address the needs of autochthonous ethnolinguistic groups whose musical heritage is poorly represented in national and commercial catalogues. These initiatives prioritise community stewardship, cultural sensitivity, and ethical governance over scale, and illustrate how diversity-aware data infrastructures must remain adaptable to the capacities and values of the communities they serve. Conversely, the central modules of the Open Music Observatory are designed to operate at a supra-national, pan-European level, supporting comparative analysis, research, and policy coordination without subsuming national or community-level autonomy. Taken together, these layered initiatives illustrate a coherent architectural approach grounded in subsidiarity: community-curated data spaces for small ethnolinguistic groups, nationally governed reference databases for policy implementation, and federated Europeanlevel modules for aggregation and comparison. Within this framework, SKCMDb functions as a national node that connects local cultural realities to European policy objectives. Rather than treating policy compliance as an enforcement problem, the approach reframes it as a data infrastructure problem: how to compute eligibility and diversity indicators in a manner that respects privacy, artistic autonomy, and institutional realities, while remaining interoperable across levels of governance. This paper proceeds as follows. Section summarises the findings of the feasibility phase that motivated the work. Section situates the project within the European data governance agenda. Sections Section and Section describe the methodological foundations and the separation of internal evaluation from external disclosure. Section builds on the privacy aspectes of Section and presents the federated implementation and system architecture. Section demonstrates the approach through a case study based on official Slovak music charts. The paper concludes by discussing implications for European cultural policy, data governance, and the future development of the Open Music Observatory as a prototype European Music Observatory. Findings of the Feasibility Phase The Feasibility Study On Promoting Slovak Music In Slovakia & Abroad4identified a persistent and structural gap between the normative ambitions of cultural policy and the operational realities of music dissemination. While local content quotas and diversity objectives are typically articulated in legal or institutional terms, they are rarely accompanied by a technical or methodological framework that would allow their consistent application in 4The study was published in English as Feasibility Study On Promoting Slovak Music In Slovakia & Abroad and Projektová štúdia propagácie slovenskej hudby na Slovensku a v zahranicí in Slovak (Daniel Antal 2020a, 2020b). 5 everyday editorial practice. This gap was not merely practical, but conceptual: policy definitions often relied on categories that could not be coherently mapped onto the objects actually circulated by media platforms. A central finding concerned the mismatch between persons and cultural artefacts. Radio stations, streaming platforms, and broadcasters do not distribute composers or performers as such; they distribute recorded fixations of performances, that is, sound recordings. These recordings are cultural artefacts that do not themselves possess personal characteristics. Yet many quota systems implicitly assume that attributes of persons—such as nationality, age, gender identity, ethnicity, or place of residence—can be directly ascribed to the recordings through which music is communicated to the public. This assumption breaks down both conceptually and legally. Modern European cultural policy does not follow nineteenth-century ethnomusicological traditions that attempted to define what makes music intrinsically “Slovak”, “Hungarian”, or otherwise national. Contemporary quota systems instead operate through agent-based criteria: a recording may qualify as local or national if it is a performance of a musical work composed by a person born in, residing in, or otherwise connected to a given country; in other cases, language of lyrics or place of performance may be decisive. In Slovakia, for example, vocal recordings performed in the Slovak language may qualify irrespective of the performer’s place of birth, recording location, or ethnicity. A recording of Slovak folk music performed in Hungary by members of the Slovak ethnolinguistic minority may therefore legitimately count toward both Hungarian and Slovak quota systems. At the same time, European quota systems are constrained by fundamental legal principles, including the prohibition of discrimination based on ethnicity, nationality, or other protected characteristics. Cultural policy may promote linguistic diversity, local cultural ecosystems, or underrepresented groups, but it cannot lawfully categorise or exclude works on the basis of ethnic identity. This legal constraint is not a weakness of European quota systems; rather, it reflects a normative commitment to pluralism and equal treatment. It also opens space for precisely the kinds of cross-border and minority cases illustrated by Slovak folk music performed in Hungary or Hungarian-language folk traditions maintained by Csángó communities in Romania. The feasibility work was strongly influenced by insights from intersectional data feminism, particularly the recognition that data-related inequalities do not arise along a single axis, but through the accumulation and interaction of multiple structural asymmetries. In the global music industry, the visibility advantage of an American male star cannot be meaningfully compared to the position of a Slovak female composer, nor to that of a Csángó woman singing in a distinct Hungarian dialect in Romania. These cases sit at the intersection of linguistic marginality, gender imbalance, ethnic minority status, and unequal access to global distribution infrastructures. Treating any one of these dimensions in isolation risks reproducing, rather than mitigating, existing inequalities. Importantly, this intersectional perspective also revealed that diversity-oriented policies must not be framed as opposing “majority” and “minority” interests in simplistic ways. Most 6 European national music ecosystems—regardless of internal gender or ethnic hierarchies— are themselves structurally marginal within the global music economy. From this perspective, the protection of cultural diversity includes not only historically marginalised groups, but also majority male performers operating within small national markets whose repertoire competes with overwhelmingly dominant global catalogues. A European diversity framework must therefore protect all participants in fragile cultural ecosystems, while remaining attentive to internal power asymmetries. These observations further complicate the already problematic linkage between personal attributes and recordings. A single sound recording may involve multiple composers, lyricists, performers, arrangers, and producers, each situated differently across axes of gender, language, ethnicity, nationality, and residence. Group identities are fluid, and personal circumstances change over time. Attempting to stabilise these attributes at the level of the recording risks both conceptual incoherence and ethical overreach, particularly under data protection law. The feasibility study therefore concluded that sustainable implementation requires a decoupling of internal evaluation from external disclosure, informed by an intersectional understanding of power and inequality. Rather than exposing personal attributes or identity categories, a system must be able to evaluate eligibility internally—within a legally justified, institutionally competent, and ethically governed environment—and expose only the outcome of that evaluation at the level of the artefact. In this model, a recording may be labelled as meeting or not meeting a quota requirement, without revealing which contributor, which protected characteristic, or which intersection of attributes triggered eligibility. This conclusion marked a decisive reframing of the problem. Local content and diversity policies could no longer be treated as matters of editorial judgement alone, nor as enforcement exercises aimed at individuals. Instead, they emerged as intersectional information design problems: how to construct reference infrastructures that translate policy rules into computable properties of cultural artefacts, while respecting non-discrimination, data protection, institutional competence, and the complex social realities of European cultural production. This insight directly informed the subsequent architectural and governance choices underlying the Slovak Comprehensive Music Database and its federated extensions. Policy Context and the European Data Governance Agends The implementation of the Slovak Comprehensive Music Database (available at https://hudobnadatabaza.sk/), HUNDB (available at ), the Finno-Ugric Data Sharing Space (available at https://finnougric.net/) coincided with a broader shift in European policy thinking on data governance, artificial intelligence, and cultural infrastructures. Over the past decade, it has become increasingly clear that Europe’s capacity to pursue cultural, social, and economic objectives in the digital domain depends less on isolated regulatory instruments and more on the availability of trustworthy, interoperable data 7 infrastructures. This shift is reflected in the European Data Strategy, the Data Governance Act5, and the growing emphasis on sectoral data spaces as instruments of public policy. In the cultural and creative sectors, this transformation has been driven in part by the rapid platformisation of music distribution and discovery. Recommendation systems, analytics pipelines, and increasingly generative AI models now mediate access to cultural content at a global scale. These systems rely on metadata that is often incomplete, inconsistent, or governed by commercial incentives that do not align with public-interest objectives. As a result, cultural policy goals—such as diversity, linguistic plurality, gender balance, or local content promotion—are difficult to articulate, let alone enforce, within algorithmic environments dominated by non-European actors. Against this backdrop, diversity quotas and local content regulations can no longer be understood as standalone cultural policy tools. Instead, they function as stress tests for Europe’s data governance ambitions. If a policy goal cannot be expressed in machine-readable terms, evaluated using auditable data sources, and implemented without violating fundamental rights, it risks becoming symbolic rather than effective. The monitoring and enforcement gaps identified in earlier quota systems thus foreshadowed a more general problem: the absence of shared reference infrastructures capable of mediating between policy intent, data protection law, and algorithmic practice. European institutions have increasingly recognised this challenge. Rather than advocating for centralised databases or uniform metadata schemas, recent policy documents emphasise federation, interoperability, and governance-by-design. Data spaces are envisaged as socio-technical arrangements in which multiple actors retain control over their data, while agreeing on shared semantics, interfaces, and rules of engagement. This approach is particularly well suited to the cultural sector, where data is distributed across public institutions, collective management organisations, private platforms, archives, libraries, and community initiatives, each operating under different legal and ethical constraints. Within this framework, the principle of subsidiarity acquires renewed significance. Cultural identity, heritage, and creative production are deeply contextual and cannot be meaningfully governed through purely centralised mechanisms. At the same time, national solutions alone are insufficient in a digital environment characterised by cross-border circulation, global platforms, and transnational audiences. European data governance therefore requires multi-level infrastructures that operate simultaneously at community, national, and supra-national scales, without collapsing these levels into a single point of control. In the Slovak context, this governance logic was formalised through a Memorandum of Understanding between the Ministry of Culture, its analytical unit (IKP), and the OpenMuse consortium, which later served as the basis for extended cooperation with the Slovak Music Centre, the Music Fund, and eventually the Slovak National Library. The MoU does not regulate data transfer or policy authority; instead, it establishes a framework for open policy analysis, methodological transparency, and non-exclusive institutional cooperation, 5Regulation (EU) 2022/868 of the European Parliament and of the Council of 30 May 2022 on European data governance and amending Regulation (EU) 2018/1724 (Data Governance Act) (European Parliament and Council 2022) 8 allowing each participating institution to retain its legal competences and data stewardship responsibilities6. In Hungary, a comparable level of centralised coordination could not be achieved at the outset. As a result, the Hungarian implementation followed a deliberately bottom-up trajectory. Building on the Slovak template and its early public articulation in conference presentations, the Hungarian work progressed through a series of bilateral agreements, exploratory workshops, and open professional discussions with individual institutions. This process involved sustained dialogue with archival, heritage, and research actors, rather than a single coordinating authority. While slower and more fragmented, this bottom-up approach proved well suited to the Hungarian institutional landscape and ultimately produced a governance structure aligned with local capacities, traditions, and trust relationships. In this sense, the divergence between the Slovak and Hungarian governance paths is not a deviation from the shared architectural vision, but an expression of the subsidiarity principle in practice7. A comparable but more radical application of this logic underpins the Finno-Ugric Data Sharing Space, which includes community-curated databases for Livonian and Karelian musical heritage. In these cases, the institutional depth of a nation state—such as that available in Slovakia or Hungary—is largely absent. Livonian and Karelian heritage is preserved by a small number of dedicated regional GLAM institutions, research centres, NGOs, and community organisations, including the Livonian Institute. Under these conditions, a centrally coordinated governance model would be neither feasible nor appropriate. Instead, the implementation necessarily resembles community-driven approaches familiar from infrastructures such as Wikidata, MusicBrainz, or Discogs, albeit combined with more explicit data curation and governance safeguards. Authority is distributed, participation is voluntary, and trust is built through transparent methods rather than formal mandate. From a policy perspective, the Finno-Ugric Data Sharing Space functions as a subsidiarity and epistemic-justice stress test for European data governance concepts. Unlike the Slovak or Hungarian cases, it operates largely outside state-scale cultural infrastructures and engages communities whose heritage has historically been documented, classified, and interpreted through external—often post-imperial—institutional frameworks. This makes it particularly relevant for assessing whether European principles such as federation, nonextractive reuse, governance-by-design, and respect for cultural and linguistic diversity can be operationalised in low-scale cultural ecosystems. At the same time, the turn toward Finno-Ugric heritage was motivated by a pragmatic methodological consideration. While the Slovak and Hungarian implementations aimed for broad coverage across hundreds of thousands of works and recordings, such scale makes it 6Memorandum o porozumení o využití výsledkov analýz otvorených politík v kontexte slovenského kultúrneho a kreatívneho priemyslu a sektorových verejných politík v spolupráci s konzorciom pre výskum a inovácie s názvom OpenMuse. [Memorandum of Understanding on utilizing the Open Policy Analysis results of the OpenMuse Research and Innovation Consortium in the context of Slovak cultural and creative industries and sectors’ public policies] (Ministerstvo kultúry SR and Open Music Europe 2023) 7Presented at the annual conference of library and GLAM information management conference, the A szlovák adatkicserélési tér magyarországi föderációjának lehetőségei (Daniel Antal 2024c) was followed by numerous bilateral and open professional forums. 9 A comparable archival orientation characterised work with Livonian materials in Latvia, in collaboration with the Latvian Archives of Folklore. In both Hungary and Latvia, epistemological justice and archival interoperability emerged as decisive methodological concerns. Large portions of musical heritage had not been publicly distributed during periods of Soviet occupation, making the “source” of metadata repair not deficient commercial catalogues but historically constrained archival records. These contexts provided demanding stress tests for both the Heritage Digital Twin Ontology and the broader data-sharing approach, particularly in relation to rights statements and differentiated access regimes. Figure 1: Our approach was presented and discussed on DNBH 2025 in Tartu. Methodologically, the Finno-Ugric work extends this archival orientation beyond state and institutional archives into community-anchored reconstruction workflows. Rather than treating archives as authoritative endpoints, the approach treats them as starting points for semantic repair, multilingual enrichment, and participatory validation. Methodologically, the Finno-Ugric work extends this archival orientation beyond state and institutional archives into community-anchored reconstruction workflows. Rather than treating archives as authoritative endpoints, the approach treats them as starting points for semantic repair, multilingual enrichment, and participatory validation. This logic is elaborated in the Finno-Ugric Data Sharing Space preprint (Dániel Antal et al. 2025), where Wikibase-based federation, lightweight ontological patterns, and community-driven annotation are used to reconnect dispersed collections without flattening local epistemologies. In this sense, the Finno-Ugric module does not introduce a separate methodology, but represents the most constrained and therefore most revealing application of the same modular, provenance-driven approach developed in the Slovak and Hungarian cases. 16 Finally, the pan-European central module of the Open Music Observatory introduced an additional methodological requirement. At this level, the objective is not detailed curation of individual works or recordings, but aggregation and comparison across countries, repertoires, and policy regimes. This necessitated the development of abstract statistical vocabularies of diversity capable of supporting comparative analysis without hard-coding identity categories or collapsing local meanings. A dedicated case study is introduced in a later section to illustrate how these aggregated vocabularies are instantiated in practice. Taken together, these implementations demonstrate that methodological coherence does not require uniformity. On the contrary, the robustness of the overall approach depends on its capacity to adapt to different institutional settings—libraries, archives, community collections, and pan-European observatory functions—while maintaining shared principles of interoperability, provenance, and ethical governance. The following section examines how these methodological commitments enable the evaluation of diversity goals without exposing sensitive personal data. Methodological choices were further informed by sustained engagement with open knowledge and semantic web communities, particularly within the Wikimedia ecosystem. At a methodological level, several of the modelling and governance choices described above were iteratively validated through structured engagement with European research infrastructures, open knowledge communities, and sectoral expert forums. Early alignment focused on semantic modelling and ontological compatibility. In October 2023, a dedicated work session convened by the Polifonia project brought together stakeholders including Podiumkunst.net, the Open Music Observatory, the University of Florence, IC Fonseca School, Joséphine Simonnot/PRISM, Maria Luisa Onida/D’Istruzione Superiore Leonardo Da Vinci, the Carnegie Hall Archive, and the Municipality of Bologna. This consultation played a formative role in refining the modular ontology strategy adopted here13, helping to ensure compatibility between scholarly music-knowledge graphs and policy-oriented interoperability requirements, while avoiding premature ontological closure. At this stage, we started to record user stories for the design of system competences of the Open Music Europe software ecosystem and the observatory’s semantic architecture14. Subsequently, attention shifted from semantic alignment to questions of governance, federation, and data sovereignty. These issues were discussed in a broader cross-sectoral context at Dataweek²� in Leuven, within the framework of the Big Data Value Association and GaiaX15. Presenting the Open Music Observatory and its national modules as an example of a European music dataspace allowed the methodological approach to be evaluated against emerging governance expectations for federated data infrastructures, including principles 13 The Polifonia Ontology Network: Building a Semantic Backbone for Musical Heritage (Berardinis et al. 2023) and Engineering Ontologies with Patterns – The eXtreme Design Methodology (Blomqvist, Hammar, and Presutti 2016) 14The user stories were placed in an OPA compliant folder on GitHub. This work was not carried on after the initiated grant agreement change, when the WP4 did no longer participate in building open source software for this purpose, but we did not abandon the competency requirements (Open Music Europe Consortium 2023). 15Jun 5, 2024, Dataweek²�, Leuven, Belgium. 17 of subsidiarity, non-extractive reuse, and institutional autonomy. Feedback from this forum directly informed the articulation of the federated data-sharing model described in the present paper. Our work was presented in the Technology sessions of the Wikimedia CEE Meeting 2024 (Istanbul) and the Wikimedia CEE Meeting 2025 (Thessaloniki), as well as at Wikidata Conf 2025 (online) (Daniel Antal 2024a, 2025c; Daniel Antal, Pigozne, and Federico 2025). These forums provided a practical testbed for validating modelling choices, editorial workflows, and governance assumptions against large-scale, community-maintained knowledge infrastructures. Through this engagement, we established working relationships with several national Wikimedia chapters and with the Wikidata and Abstract Wikipedia teams, and participated in the Wikidata Ontology Cleanup Task Force and the Wikidata Mereology Task Force. This involvement helped align our ontological patterns, identifier practices, and documentation strategies with broader open knowledge standards, while also ensuring that our approach to provenance, uncertainty, and institutional responsibility remains compatible with community-driven curation environments. Diversity Goals Without Exposure of Sensitive Data A defining feature of the Slovak Comprehensive Music Database is the strict separation between internal evaluation and external disclosure. While the system may internally process attributes that are relevant for diversity or local content criteria, it exposes only binary or aggregated outcomes. At the level of the recording, a track is reported as meeting or not meeting a quota requirement, without revealing which contributor, which personal attribute, or which specific criterion triggered eligibility. This separation is central to the viability of diversity-aware cultural policy under European legal and ethical constraints. It operationalises the principle of data minimisation by ensuring that only the information necessary for editorial, regulatory, or analytical purposes is disclosed. At the same time, it preserves the ability to evaluate complex quota definitions that depend on personal attributes which cannot lawfully or ethically be exposed. Implementing this distinction requires more than a technical design choice. It is precisely the challenge addressed by legal, organisational, and semantic alignment under the European Interoperability Framework. The system must be able to lawfully access authoritative data sources, design an organisational workflow that produces a determinate evaluative outcome, and communicate that outcome back into a shared data space using semantics that can be consistently interpreted by diverse actors, including broadcasters, curators, researchers, and media authorities. In the Slovak context, this workflow is deeply intertwined with rights management. The national data sharing space is organised around SOZA, the collective management organisation with legal competence to identify musical works and connect them to living rightsholders and their heirs. Personal characteristics of composers—such as place of birth, nationality, residence, or ethnicity—are strictly protected under the General Data Protection Regulation 18 and are not accessible for disclosure. As a result, while the evaluative logic is conceptually straightforward, its practical implementation is necessarily mediated by institutional trust and competence. At a high level, the internal evaluation proceeds through a sequence of controlled matches. A sound recording is identified and linked to the musical work it embodies; the work is associated with its composer or composers; and the system evaluates whether any of the relevant personal or work-level characteristics satisfy the applicable quota definition. If at least one criterion is met, the recording is marked as eligible for the purposes of the local content quota. Crucially, the reason for eligibility is never exposed. A composition may qualify because the composer was born in Slovakia, because the composer resides there, or because the lyrics are written in the Slovak language. If any of these conditions are satisfied, all recordings of works by that composer—including co-authored works—are treated as eligible, without disclosing which condition applied. The result of this internal evaluation is exposed through a simple property at the level of the recording. For example, a recording may be labelled as heritage of Slovakia when it qualifies under the national quota, or as heritage of Slovaks when eligibility is based on linguistic intelligibility for Slovak speakers as an ethnolinguistic group. This modelling choice deliberately abstracts away from individual identity attributes while retaining policyrelevant meaning. It allows quota evaluation to operate at the level of cultural artefacts rather than persons. Language of lyrics occupies a particular position within this framework. Unlike personal attributes, language is an audible and editorially relevant characteristic of a recording, yet it remains culturally mediated rather than strictly objective. In SKCMDb, lyrics are treated as Slovak when they are intelligible to a native Slovak speaker, even if they are performed in dialectal or archaic forms, as is common in folk traditions. Such determinations may be established by native-speaker curators, performers, or composers. This approach recognises linguistic variation without essentialising identity or excluding historically rooted forms of expression. The is heritage of property used to express these outcomes is defined in the ECCCH Heritage Digital Twin Ontology, which subclasses CIDOC CRM and is widely used in museum and heritage contexts. In SKCMDb, this property—also used under the alias local to—represents a form of community stewardship governed primarily by ethical frameworks such as CARE, rather than by legal entitlement alone. In the Hungarian replication, the same property is used to refer not only to Hungarians as a national group, but also to linguistically distinct communities such as Székely or Csángó speakers in present-day Romania. A Csángó folk song may thus be treated as part of the Hungarian national quota for policy purposes, while retaining a more precise ethnolinguistic designation for ethnomusicological analysis. The same abstraction is applied at the level of nation-states. A recording is treated as heritage of or local to Slovakia or Hungary if it meets the respective national quota regulations. This modelling choice enables the computation of statistical indicators—such as the proportion of locally relevant repertoire in charts or playlists—without exposing personal data or identity categories. It also allows the same underlying evaluation logic to support multiple 19 policy objectives, including radio and television quotas, festival programming diversity, and international initiatives such as gender equality pledges. It is important to note that the impact of the General Data Protection Regulation on the cultural heritage, music, and digital humanities sectors extended far beyond the specific context of diversity quotas. While GDPR contains differentiated provisions for libraries, archives, and other GLAM institutions, its introduction nevertheless came as a structural shock to the sector. Long-standing practices of metadata publication, authority sharing, and collaborative documentation—often developed in good faith and under earlier legal regimes—suddenly became legally ambiguous or non-compliant. This applies in particular to retroactively affected archival materials and to cross-institutional authority initiatives, such as shared name files or federated registries, which rely on the controlled exchange of personal data. Within this context, the architecture developed for SKCMDb supports not only policy evaluation, but also institutional compliance and remediation. By enabling the lawful use of GDPR-protected data within trusted environments to correctly link agents to archival, library, museum, and rights registry holdings, the system helps participating institutions to repair metadata, clarify authorship and neighbouring rights relationships, and address potential moral rights issues. Crucially, this work remains internal to competent institutions, while its outcomes—such as corrected links, clarified work–recording relationships, or eligibility evaluations—can be safely exposed at an abstracted level. In this sense, the approach taken here does not merely limit disclosure to avoid legal risk. It creates the conditions under which heritage institutions may, after careful analysis, increase the educational, research, or public visibility of their holdings in a legally and ethically defensible manner. Rather than treating GDPR as an external obstacle to cultural data work, the system incorporates it as a design constraint that, when properly addressed, enables more responsible sharing, clearer rights attribution, and more sustainable interoperability across the music and heritage sectors. By design, this approach prevents the system from becoming a tool for profiling or surveillance. Editors and analysts receive determinate, interpretable results, while the underlying personal data remains protected within competent institutions. In this way, diversity goals are operationalised not through the exposure of identity, but through the careful design of evaluative workflows that respect legal constraints, ethical commitments, and the complex social realities of European cultural production. Application Layers and System Architecture The layered system architecture of SKCMDb is a direct consequence of the federated governance model described in the preceding sections. The separation between data stewardship, interoperability, public access, and analysis is not merely a technical design choice, but an operationalisation of institutional competence, legal responsibility, and subsidiarity. In particular, it reflects the need to support policy-relevant evaluation and analytical reuse while ensuring that sensitive data remains under the control of the institutions legally entitled to process it. 20 At the foundation of the architecture is a private or trusted layer, operated by institutions with formal legal competence over authoritative data. In the Slovak case, this layer is anchored in the systems of the collective management organisation SOZA and, where relevant, other trusted institutional partners. This layer contains detailed information on musical works, rightsholders, and personal attributes that may be necessary for internal evaluation, rights clearance, or metadata repair, but which are protected under data protection law. Access to this layer is governed by contractual, legal, and organisational safeguards, reflecting the governance arrangements established through the national Memorandum of Understanding described in Section . Its contents are never exposed directly to external users or applications. Above this sits an intermediate interoperability layer, which functions as a controlled interface between trusted data holders and the wider data sharing space. In the SKCMDb implementation, this role is fulfilled by a SPARQL endpoint implemented using Apache Jena Fuseki. This layer hosts derived RDF graphs generated from authoritative sources and refreshed on a regular basis. Crucially, these graphs are treated as reproducible publication artefacts rather than systems of record. They contain only non-personal data and outcome-level properties resulting from internal evaluation, such as eligibility flags or heritage associations. In this way, the interoperability layer translates institutionally governed microdata into a form that can circulate safely across organisational and national boundaries without exposing sensitive underlying information. The public access layer builds on this interoperability interface to support exploration, navigation, and reuse by a broader audience. In several modules of the Open Music Observatory ecosystem, including the Slovak implementation, this layer is realised through Sampo-style user interfaces. These interfaces are designed primarily for cultural heritage, research, and policy contexts. They allow users to browse repertoires, explore relationships between works and recordings, and apply faceted filters based on non-sensitive attributes. From a governance perspective, this layer embodies the commitment to transparency and public value articulated in the Slovak MoU, while remaining firmly decoupled from any processing of personal data. The analytical layer provides tools for computation, aggregation, and visualisation, and is explicitly constrained to operate on the same outcome-level data exposed through the interoperability layer. In the Slovak case, this includes applications built with Shiny, which enable editors, analysts, or researchers to upload playlists or chart data and compute diversity or local-content indicators based on matches against the SKCMDb graph. These analytical tools do not require, and do not permit, direct access to trusted institutional data. Their design reflects the separation between internal evaluation and external use described in Section ensuring that policy-relevant analysis can be performed without expanding the circle of access to protected data. This layered architecture also enables replication across jurisdictions and institutional contexts. A similar configuration can be deployed elsewhere by identifying appropriate data stewards for the trusted layer, exposing derived graphs through a standardised interoperability interface, and reusing or adapting existing public and analytical applications. Because the analytical logic operates on outcome-level properties rather than raw personal data, the same tools can be reused across national implementations even where underlying legal 21 regimes, institutional arrangements, or data sources differ. This was precisely the case in the Hungarian implementation, where a bottom-up governance model required tighter institutional boundaries, and in the Finno-Ugric data spaces, where trusted layers are distributed across NGOs and regional GLAM institutions rather than central authorities. Importantly, the architecture also supports asymmetry between modules. Not all implementations are required to expose the same interfaces or support identical analytical workflows. In contexts where statistical monitoring is a priority, such as national broadcast quotas, the analytical layer may focus on time-series aggregation and indicator computation. In archival or ethnomusicological contexts, by contrast, public interfaces and analytical tools may prioritise qualitative exploration, provenance tracing, or competence questions that are not immediately reducible to numerical indicators. The architecture accommodates these differences without fragmentation because all layers remain anchored in shared semantic models and governance principles. By deriving its technical structure from governance and policy constraints rather than imposing them retroactively, the SKCMDb architecture ensures that diversity-aware cultural policy objectives can be supported without compromising data protection, institutional autonomy, or ethical commitments. At the same time, it creates a clear and auditable pathway from authoritative microdata to aggregated indicators. This makes it possible to demonstrate the system’s capabilities through concrete empirical applications. The following section presents such an application, applying the SKCMDb infrastructure to the computation of local content indicators derived from official Slovak music charts. Case Study: Computing Local Content Indicators from Official Charts This section presents an empirical application of the SKCMDb infrastructure to demonstrate how diversityand locality-related indicators can be computed reproducibly, lawfully, and without exposing sensitive personal data. The case study applies the methodological, governance, and architectural principles developed in the preceding sections to an authoritative, real-world dataset: the official Slovak music charts. Its purpose is not to evaluate individual actors or enforce compliance, but to show how a shared reference infrastructure can support informed, voluntary, and legally robust engagement with cultural policy objectives. Data sources and scope The empirical material for this case study consists of 156 consecutive weeks of the official Slovak Top 50 charts, covering a three-year observation period. The charts were provided directly by IFPI ČR/SR, the recognised owner and administrator of the official Slovak charts. This ensures that the dataset reflects the repertoire actually used for public reporting, industry benchmarking, and broader discussion of music consumption in Slovakia. All charting entries across the observation period were ingested into the Slovak Comprehensive Music Database. Rather than sampling or pre-filtering the data, the case study deliberately includes the full set of charting recordings. This choice serves two purposes: it 22 demonstrates the scalability of the approach, and it avoids biases that could be introduced through selective inclusion or editorial filtering. Rationale for the use of official charts as a neutral reference corpus The choice of official music charts as the empirical basis for this case study was deliberate and reflects both legal and institutional considerations identified during the feasibility phase. In Slovakia, local content regulations apply directly to radio broadcasters, who are legally required to meet minimum quotas for domestic repertoire. At the same time, consultations conducted as part of the Listen Local feasibility study—funded by the Slovak Arts Council and later extended within the OpenMusE project supported by the European Commission— revealed a recurring concern among broadcasters: while they are subject to regulatory obligations, they have received limited institutional support to monitor compliance in a reliable, consistent, and legally safe manner. From the perspective of editors, the “Slovakness” of a recording is often not directly observable. Audible language, artist name, or registration metadata may provide partial signals, but they are insufficient to determine eligibility under the full legal definition of the quota. This opacity creates uncertainty and can encourage overly cautious programming strategies that privilege already well-established repertoire, undermining the intended policy goals of diversity, renewal, and inclusion. One of the core aims of the feasibility work, and of SKCMDb more broadly, was therefore to provide tools that support compliance rather than enforce it. The objective is to lower the informational and organisational burden on editors and curators, enabling them to make informed decisions without exposing personal data or assuming regulatory authority. This support-oriented approach extends beyond licensed broadcasters. Local bars, hotels, cultural venues, and independent playlist curators—many of whom are not formally subject to quota obligations—may nevertheless wish to contribute voluntarily to local cultural ecosystems. In cities such as Košice, for example, there is reason to believe that many venues would be willing to give greater visibility to local artists if they could be confident that public performance fees and cultural attention flow back into the local music community. For these reasons, the case study deliberately avoids analysing the playlists or programming decisions of individual radio stations. Instead, it uses the official Slovak charts as a neutral reference corpus. The charts are widely recognised, industry-authoritative, and relevant to all broadcasters, yet they do not belong to any single station. This choice avoids any implication of monitoring or evaluating specific actors, while still allowing the demonstration of how diversity and locality indicators can be computed from shared, publicly acknowledged reference data. By grounding the empirical application in a neutral dataset, the case study positions SKCMDb as a common infrastructure serving the entire ecosystem. It demonstrates how authoritative reference data can be transformed into policy-relevant indicators that are usable by broadcasters, venues, and curators alike—without creating new asymmetries of power or shifting regulatory responsibilities onto the data infrastructure itself. 23 Microdata layer: recordings and works Each chart entry was first resolved to a sound recording, identified by standard industry identifiers where available (such as ISRC), complemented by platform identifiers and curated matches where necessary. Recordings were then linked to the musical works they embody. This step is essential, as eligibility for local content quotas in Slovakia depends in part on attributes of the composer, which are properties of the work rather than of the recording or the performer. Within SKCMDb, both recordings and works are represented as Wikibase items. In methodological terms, each item functions as a structured dataset consisting of statements, qualifiers, and provenance information. This representation allows the system to accommodate uncertainty, multiple assertions, and evolving knowledge, while maintaining traceability and institutional accountability. The matching process relies on authoritative sources mediated through trusted institutional partners, particularly collective management organisations. Where automated matching is possible, explicit identifiers are preferred; where human curation is required, decisions are recorded with provenance. This conservative approach reflects the methodological stance articulated earlier: correctness and accountability take precedence over maximal coverage. Internal evaluation and indicator logic For the purposes of this case study, two indicators were defined in line with Slovak regulatory and editorial practice: 1. Slovak-language share, indicating whether a recording qualifies based on the intelligibility of its lyrics for Slovak speakers. 2. Slovak-origin share, indicating whether a recording qualifies based on the eligibility of its composer or composers under national criteria, such as residence or other legally recognised connections. The internal evaluation logic follows the principles described in Section 5. For each recording, the system evaluates whether any of the applicable criteria are met. If at least one criterion applies, the recording is marked as eligible. Crucially, the system does not record or expose which specific criterion triggered eligibility, nor which individual contributor satisfied it. The outcome is expressed solely as an artefact-level property of the recording. This logic allows complex, legally constrained definitions of locality to be operationalised without exposing personal data or identity categories. It also ensures consistency across recordings by the same composer, including co-authored works, while avoiding person-level disclosure. 24 Aggregation and observatory-level representation Eligibility outcomes were aggregated at two temporal resolutions: weekly, corresponding to individual chart publications, and annual, corresponding to calendar-year summaries. For each period, the proportion of chart entries qualifying under the Slovak-language and Slovak-origin indicators was computed. These aggregated results were then expressed as statistical data cubes within the central modules of the Open Music Observatory. Time was modelled explicitly as a dimension, with measured values representing the respective shares. By lifting the results from item-level evaluations into a statistical representation, the case study demonstrates how national-level reference data can contribute to supra-national observatory functions without transferring underlying microdata or sensitive attributes. Importantly, the observatory-level representation contains no information that would allow the reconstruction of personal characteristics or editorial decisions at the level of individual works or contributors. What circulates upward is a set of interpretable indicators, not the data used to compute them. Interpretation limits and policy relevance The purpose of this case study is not to assess compliance by individual broadcasters, nor to evaluate the effectiveness of Slovak cultural policy. The indicators presented are descriptive, not normative, and they do not imply causality. Their function is to demonstrate that diversityand locality-related metrics can be computed in a manner that is auditable, reproducible, and compatible with European data protection and non-discrimination principles. From a policy perspective, the case study illustrates how authoritative industry data, rightsmediated evaluation, and federated data governance can be combined to support evidenceinformed discussion without creating new surveillance or enforcement mechanisms. From a methodological perspective, it shows that the same infrastructure can support both finegrained microdata curation and high-level statistical aggregation, depending on the level at which questions are asked. In this sense, the charts analysis serves as a proof of implementation rather than as a standalone analytical result. It demonstrates that the SKCMDb architecture is capable of translating abstract policy criteria into computable indicators, while preserving institutional autonomy, protecting sensitive data, and remaining interoperable within a European data space. Discussion and Outlook This paper set out to address a persistent gap between the ambitions of diversity-oriented cultural policy and the practical realities of its implementation in contemporary music 25