The first group comprises the most relevant studies, which have already illuminated specific aspects of interoperability between Schema.org and RDA. Senior (2018) examined possible alignments between key elements likely to be adopted for serial description from emerging metadata standards, including Schema.org and the RDA Registry¹. By mapping specific classes and elements, this study revealed that while interoperability exists in the mapping of identifiers and chronology, significant challenges persist due to conceptual differences, Schema.org’s multiple-class inheritance hierarchy, and its relatively limited granularity in serial description. Similarly, Fardehosseini (2019), as part of her doctoral research, conducted a qualitative content analysis to map Schema.org properties to RDA elements. The study concluded that although Schema.org properties can generally be mapped to RDA elements, RDA offers more comprehensive elements for describing bibliographic entities. Notably, neither of the studies incorporated RDA guidelines, instructions, or VESs into their analyses, thereby leaving the question of RDA’s potential to serve as Schema.org’s complementary content standard entirely open for future investigation.

In a less relevant study of this group, Park and Kipp (2019) examined the mapping of elements adopted by diverse libraries from different metadata standards, including Schema.org and RDA. Their findings highlighted that interoperability issues stem from differences in underlying data models—a challenge also present in the case of Schema.org and RDA, as elucidated by Senior (2018) and Fardehosseini (2019).

This particular challenge suggests the ineffectiveness of using RDA elements as intermediary components for examining the alignment inquired in the present study. Consequently, it would also complicate the analysis of the Official post-3R RDA² due to its element-based presentation of value-assignment components. This study used the Original RDA Toolkit³ instead to avoid this complication. This methodological adjustment also provided the study with a stable baseline—consistent with current metadata creation practices—and ensured comparability with existing metadata profiles, given the ongoing reliance of institutions on the Original RDA⁴. However, future research should investigate the implications of the official version to support institutions in navigating the forthcoming shift.

Similarly, although considering alignments between the entity types of the standards could have benefited the study, the absence of an official mapping undermined its applicability. Developing an alignment tailored to the present study, although an option, was unlikely to yield a conclusive outcome due to the fundamentally distinct bibliographic entity typologies of Schema.org and RDA. One pertinent example is the opposition of RDA’s WEMI (Work, Expression, Manifestation, Item) approach to Schema.org’s creative works typology, which is based on bibliographic content types and carrier types (Fardehosseini, 2019; Senior, 2018).

The second group includes studies that, similar to the present study and the previous group, contribute to the usability of Schema.org for describing bibliographic entities. The most relevant studies in this group are those concerning the association between Schema.org and other bibliographic standards, models, or ontologies. Apparent as the earliest study of this type, Godby (2013) investigated the relationship between BIBFRAME and OCLC’s linked data model of bibliographic description derived from Schema.org. Expanding on this, Godby and Denenberge (2015a) analyzed the compatibilities between the linked data models of the Library of Congress and OCLC. In other research, Nurmikko-Fuller et al. (2016) and Jett et al. (2016) compared various bibliographic metadata ontologies, including Schema.org, for specific contextual purposes. A more relevant study by Fardehosseini et al. (2020) employed qualitative content analysis to improve the functionality of Schema.org properties by representing them in the Library Reference Model (LRM). They highlighted the ontological differences between Schema.org and LRM, highlighting Schema.org’s limitations in representing relationships between bibliographic entities. They concluded this study by underscoring the role of LRM-based standards, such as RDA, in improving Schema.org’s bibliographic functionality. Further contributions to this domain are provided by Fardehosseini et al. (2021), who examined the correspondence between Schema.org’s entity types and MARC’s genre terms, Wu et al. (2023), who analyzed crosswalks from fourteen research data schemas to Schema.org, and Mohammadi Ostani et al. (2024), who investigated the enrichment of Schema.org “Thesis” properties using specific cultural heritage metadata standards.

The second group may also be supplemented by studies that contribute to the same topic but do not focus on its association with other standards, ontologies, or models. As highly relevant contributions of this type, Han et al. (2015) examined Schema.org’s effectiveness in describing library holding data, with attention to the Schema Bib Extend Community Group’s (2016) recommendation; Mohammadi Ostani et al. (2021) investigated the localization of Schema.org for manuscript description in the Iranian-Islamic information context through documentary and qualitative content analysis; and, Taheri et al. (2022) studied the Search Engine’s responses to authority data properties embedded into Schema.org-based metadata on the Microdata syntax.

Despite the valuable efforts made by the studies of this group to pave the way for Schema.org into bibliographic metadata, its compatibility with bibliographic metadata content appears to remain unaddressed by prior research. The present study seeks to respond to this gap by examining RDA’s potential to enhance the compatibility of Schema.org with bibliographic metadata content.

The third group consists of studies that have focused on RDA’s association with metadata standards, metadata ontologies, or metadata models other than Schema.org. Among the most relevant studies of this group that have investigated the interoperability of RDA with one or multiple metadata standards (Patrício et al., 2025; Taniguchi, 2023; Taniguchi, 2025; Wacker et al., 2011; Zapounidou, 2020; Zapounidou et al., 2019), only Wacker et al. (2011) focused on RDA’s content standard functionality. To address research gaps concerning RDA’s interoperability with non-MARC metadata standards, Wacker et al. (2011) reported on tests conducted by three institutions that applied RDA as a content standard to their Dublin Core, Metadata Object Description Schema (MODS), or Encoded Archival Description (EAD) metadata. They concluded that the lack of established frameworks and applications specifically designed to support the integration of RDA as a content standard for non-MARC metadata limited both the scope of the tests and the ability to fully realize and apply RDA’s benefits. The present study, in contrast, adopts an a priori approach to address a similar research gap, given that Schema.org–RDA integration remains unestablished. However, future research may benefit from a posteriori approaches once this integration is fully realized and implemented by relevant institutions.

Additional contributions include Baker et al. (2014) and Seikel and Steele (2020), who comparatively analyzed multiple bibliographic models, including RDA. It is also noteworthy that in a pertinent non-research document, Dunsire (2009) discussed the application of RDA as a content standard for UNIMARC metadata, emphasizing the standards that facilitate their interoperability. Taking advantage of such emphasis seems to be ineffective in the present case, given the current contextual distance between Schema.org and RDA. LRM potentially serves as an effective harmonization medium for RDA (Dunsire, 2023; Oliver, 2021; Oury et al., 2018), but it is likely to prove less effective in the case of its interoperability with Schema.org, which follows a different ontological approach.

Beyond the three groups reviewed thus far, additional works also offer insights into semantic interoperability between metadata standards and content standards. For instance, Nilsson et al. (2006) presented a conceptual framework that distinguishes between element vocabularies and value vocabularies, suggesting the possibility of alignment between vocabularies from both standard types. A particularly relevant study by Chan and Zeng (2006) offered a detailed methodological analysis of schema-level interoperability, with implications for aligning metadata and content standards. Their emphasis on crosswalks that map not only element names but also cardinality, repeatability, and value formats informs how one might align components between metadata and content standards. By explicitly separating property semantics from value constraints, they further showed how application profiles can be constructed to “snap together” (Duval et al., 2002) chosen properties with their corresponding value-assignment components. Furthermore, their recommendation to register both structural schemas and content rule sets in a central registry suggests a way to maintain and publish the mappings as a machine-readable profile, ensuring consistency and enabling automated validation in metadata management applications. More recently, Hendrawan et al. (2024) mapped metadata content from International Standard Archival Description (ISAD) and Cataloging Cultural Objects (CCO) into Dublin Core, showcasing how content standards from different contexts (archives, museums, and libraries) can be semantically aligned through a shared metadata standard. The study highlighted that effective retrieval depends on harmonized content standards and descriptive conventions, and without compatible value structures and terminology, even well-mapped metadata may fall short in supporting cross-domain discovery.

It emerges that the existing relevant literature predominantly focuses on interoperability among metadata standards and models. The present study appears to be among the very few that examine interoperability between specific metadata and content standards (Hendrawan et al., 2024; Wacker et al., 2011). The reviewed literature underscores “mapping” as a well-established technique for both examining and achieving semantic interoperability, which may also apply to the interoperability under investigation.

Given the breadth of the Schema.org bibliographic entities and the extensiveness of their properties, purposive sampling was employed to enhance the feasibility and depth of the analysis. Regarding the objective of the present study and its applied approach, sampling decisions were guided by relevance to the description of diverse bibliographic entities. In line with this sampling strategy, properties were selected based on three levels of entity type granularity enabled by Schema.org’s property-inheritance mechanism. Thanks to this mechanism, selecting properties from top-level entity types also included fundamental properties from their subtypes, thereby facilitating the generalizability of the analysis without the need to include all properties from lower levels of the entity type hierarchy.

The first level focused on properties shared across diverse bibliographic entities—that is, properties applicable to any entity relevant to the bibliographic universe, including content objects, agents, and places. Accordingly, this level incorporated all properties defined under Schema.org’s “Thing” class, which are inherited by all other Schema.org entity types. This inclusion thereby contributed to the generalizability of the analysis across different types of bibliographic entities.

The second level involved properties shared across diverse creative works—including both their concrete and abstract instances—as entities highly relevant to the bibliographic universe. Theoretically, this high relevance is evident in various definitions and descriptions of the bibliographic universe, evaluated by Cossham (2017) in the second chapter of her PhD dissertation (pp. 14–36). Accordingly, the samples for this level were selected based on the type-specific properties⁵ of “CreativeWork”, whose practical relevance to the bibliographic universe has been recognized by Godby (2014), Godby and Deneberg (2015a), and Fardehosseini et al. (2020). Regarding the study’s scope, the properties “schemaVersion”, “sdDatePublished”, “sdLicense”, and “sdPublisher” were purposively excluded, as they apply to the description of entity descriptions (i.e., meta-metadata) rather than the entities themselves. Additionally, given the inapplicability of the type-specific properties of “CreativeWork” to item-level descriptions (Godby et al., 2015b), this level also included five “Offer” properties recommended by the Schema Bib Extend Community Group (2016) for library holding data. This level further enhanced the generalizability of the analysis across Functional Requirements for Bibliographic Records (FRBR) Group 1 entities (i.e., works, expressions, manifestations, and items), which are also incorporated into LRM and adopted by both the original and the official version of RDA (Croissant, 2012; RDA Board, 2019; Riva et al., 2018).

The third level considered properties specific to a common type of creative work. This additional level was considered solely to better reflect real-world bibliographic description practices where distinct profiles are often developed for different types of content objects (Dunsire, 2021a; Gavili Kilaneh et al., 2019). This level included the six type-specific properties of Schema.org “Book” as a well-recognized type of content object and a widely used subclass of Schema.org “CreativeWork” on the Web, according to data published by Web Data Commons (2024).

Ultimately, the sampling resulted in the selection of 12, 105, 6, and 5 properties, respectively, from Schema.org’s “Thing”, “CreativeWork”, “Book”, and “Offer” types. Due to Schema.org’s inheritance mechanism, these properties simultaneously encompass 12 properties across all Schema.org entity types, in addition to the 105 properties shared by every “CreativeWork” subclass—including “Article”, “Dataset”, “Thesis”, etc.—thereby ensuring consistent coverage across diverse types of bibliographic entities. All selected properties can be found in the tables in the Results section. The data collection described is consistent with Schema.org’s V29.3 release⁶.

To align the scope of RDA components with the research aim, purposive sampling was employed to exclude RDA guidelines and instructions not directly relevant to the description of bibliographic entities. This exclusion comprised meta-metadata components, such as those associated with “Status of Identification”, “Source Consulted”, and “Cataloger’s Notes”. Furthermore, in order to align these components with the scope of the selected properties, the exclusion was extended to specific manifestation-level guidelines and instructions that did not apply to book descriptions, as well as to guidelines with very broad applicability. This sampling yielded 12 guidelines and 266 instructions, available in Supplementary Table 1 and Supplementary Table 2.

The guidelines and instructions were systematically collected according to their hierarchical position in the Original RDA Toolkit. To facilitate their analysis, presentation, and identification, the selected scopes of instructions were cataloged using the numerical format: “[SectionNo].[ChapterNo].[MainHeadingNo]”; while guidelines, as broader components, were cataloged using the format: “[SectionNo].[ChapterNo]”. Subsequently, they were cataloged under their associated RDA Entities⁷. This association was double-checked for ambiguous cases by identifying the Entities associated with their respective properties in the RDA Registry. This clarification further revealed that the accessibility content instructions (7.14.1) were later considered for describing manifestations despite their initial reference to expressions. It should be acknowledged that the instructions applicable to the description of representative expressions were listed only under the “Expression” Entity in order to distinguish them from other instructions associated with “Work”. Ultimately, the selected guidelines and instructions were cataloged under one or more of the following RDA Entities: “Work”, “Expression”, “Manifestation”, “Item”, “Person”, “Family”, “Corporate Body”, and “Place”. The RDA Entities associated with each scope of the RDA guidelines and instructions are illustrated in Supplementary Table 1 and Supplementary Table 2.

On the other hand, RDA VESs were selected through census sampling. The primary VESs encompassed RDA and ROF—abbreviation for RDA/ONIX (Online Information Exchange) Framework—value vocabularies collected from the RDA Registry. Although ROF value vocabularies are designed for the publishing context, they were included because of their potential usability in the bibliographic context. The collected value vocabularies are consistent with the v5.3.1 release of the RDA Registry⁸.

The study also took a step further by including RDA relationship designators, which further support RDA’s content standard functionality by providing vocabularies to refine property expressions. The inclusion of these vocabularies was motivated by their potential to enhance Schema.org’s functionality in the context of bibliographic description, where relationship designators are utilized to provide descriptive depth while avoiding excessive property proliferation.

Building upon the data collection phase, one-to-many mappings were developed from the selected Schema.org properties to the selected RDA components. The objective was to identify components that potentially align with one another based on their predefined semantic specifications, including their respective names, definitions, domain entity types, expected value types, and usage examples. The critical criterion for this mapping was the applicability of RDA components in supporting value assignment for each property. One-to-many mappings were deemed sufficient, as value-assignment components are typically applied to a preselected set of properties in real-world practices.

The mappings were developed with respect to the RDA Entities associated with the components to enhance precision. Specifically, the guidelines and instructions under “Work”, “Expression”, and “Manifestation” were considered valid options for the type-specific properties of “CreativeWork” and “Book”, while those under RDA “Item” were considered for the selected “Offer” properties. On the other hand, due to their broad bibliographic entity coverage, the “Thing” properties were open to all RDA components. The mapping also acknowledged that properties whose associated RDA Entities are apparent should only accept RDA components from those Entities. For example, “publisher” as a manifestation property should only accept components applicable to the description of manifestations.

Another procedure analyzed the alignment between Schema.org properties and RDA relationship designators. First, those properties that could accept RDA relationship designators were identified from the selected properties. Subsequently, the other relevant properties were mapped to relationship designators that could apply to the initially identified properties. The primary source for these designators was the Original RDA Toolkit’s Appendices I to M. However, they were also compared against the RDA Registry vocabularies to ensure their currency.

To ensure the validity of the mappings, three complementary validation methods were employed with respect to the study’s qualitative, applied aim: (1) cross-checking with authoritative reference materials, (2) expert review, and (3) practitioner feedback.

First, the identified RDA components that potentially corresponded to the target properties were cross-checked against authoritative reference materials: RIMMF3 (Fritz and Fritz, 2019) and resources from the Program for Cooperative Cataloging (2024). Although these materials focused on RDA as a content standard for non-Schema.org metadata standards, they provided useful models and patterns for selecting appropriate value-assignment components for analogous properties.

Second, the proposed mappings were reviewed by five purposively selected knowledge organization (KO) experts who were familiar with both RDA and Schema.org. The experts were chosen to provide complementary perspectives from the relevant subfield, including (1) library metadata management & cataloging, (2) Semantic Web & Linked Data, and (3) knowledge organization standards and systems. No new conceptual issues arose after the fifth expert review, which was interpreted as evidence of conceptual saturation within the scope of the mappings.

Finally, the mappings were implemented in a real-world metadata management environment—presented later as prototypes (see Figs. 1,2,3)—used by various organizations, and feedback was solicited from five purposively selected catalogers from different institutional contexts (small/medium/large, special/academic libraries) and experience levels (2–12 years) regarding the appropriateness and practical utility of the mappings for their cataloging workflows. Despite the modest numbers, the high level of satisfaction and the diversity of feedback across these groups support the practical acceptability of the proposed mappings.

Taken together, these three validation methods formed a triangulated validation package, thereby reducing reliance on any single method and enhancing the overall robustness of the validation.