Ontological Data Sharing of Open Government Data for Data Curation/Le partage ontologique de données gouvernementales ouvertes aux fins de la conservation des données

Abstract

The purpose of this study is to inform ontological data-sharing practices in the cultural heritage community by using open government data for data curation. Open government data have grown rapidly to enhance civic engagement and the transparency of governmental authorities in many parts of the world. We used a mixed methods approach, including co-word analysis of a population of open government data followed by qualitative analysis of five carefully selected records mapped using the CIDOC CRM. A final step was to use FRBRoo, an extension of the CIDOC CRM, to map the instantiation of data records in a typical data-sharing scenario.

Résumé

Le but de cette étude est de faire connaı^tre les pratiques ontologiques de partage de données dans la communauté du patrimoine culturel en utilisant des données gouvernementales ouvertes, aux fins de la conservation. Les données gouvernementales ouvertes ont augmenté rapidement dans de nombreuses régions du monde, renforçant l'engagement civique et la transparence des autorités gouvernementales. Nous avons utilisé une approche par méthodes mixtes, dont une analyse des cooccurences de termes au sein d'une population de données gouvernementales ouvertes, et une analyse qualitative de cinq notices soigneusement sélectionnés en utilisant le modèle conceptuel de référence CRM CIDOC. La dernière étape a consisté à utiliser FRBRoo, une extension du CRM CIDOC, pour cartographier l'instanciation des enregistrements de données dans un scénario de partage de données typique.

Introduction

The purpose of this study is to inform ontological data-sharing practices in the cultural heritage community by using open government data for data curation. The research problem stems from the current rapid increase in the quantity of open government data in the cultural heritage domain. However, many cultural heritage open government data cannot be shared because those data are "stuck" in siloed, unstructured, and heterogeneous file formats, data formats, or Web addresses. For example, heterogeneous file formats range from MS Excel and HWP (Hangul Word Processor), to XML (eXtensible Markup Language), RDF (Resource Description Framework), and PDF (Portable Document Format) and to Web addresses simply linked to other open government websites. Governmental investments in making open government data available to the general public are increasing. But data sharing of cultural heritage open government data, which could create new value for the general public, has not been actively studied. Also, there has been almost no research on knowledge organization with regard to ontological data sharing using open government data. Knowledge organization as a science has recently been defined as "The evolving science of the order of knowledge, which informs the application of functional knowledge organization systems … Knowledge Organization works with shared perceptions represented by conceptual phenomena" (Smiraglia 2015b, 1).

A primary research question, then, is how to share cultural heritage open government data that might be stuck in heterogeneous and unstructured formats by using ontologies for data curation. With this study, we hope to demonstrate a knowledge organization approach to sharing open government data in the cultural heritage domain to create new accessibility for the general public.

Background

Open government data

Open government data refer to "non-privacy-restricted and non-confidential data which is produced with public money and is made available without any restrictions on its usage and or distribution" (Janssen, Charalabidis, and Zuiderwijk 2012, 309). Open government data open to the general public have grown rapidly in quantity to enhance civic engagement and the transparency of national, state, and local governmental authorities in many parts of the world, including Austria, Australia, New Zealand, the Republic of Korea, Sweden, the United Kingdom and the United States.

However, means of enhancing data sharing in the cultural heritage domain using open government data available to the general public have not been actively studied. Among cultural heritage institutions, libraries, archives, and museums usually are included. Open government data sharing has focused on structured data sets such as RDF formats (e.g., linked data). With regard to the increasing necessity of data sharing of open government data relevant to cultural heritage information, only a few studies have been conducted because of unstructured data sets that are found in varying heterogeneous formats. This paper presents one of a series of case studies in which open government cultural heritage data records are mapped for data sharing using a metalevel ontology.

Data curation

Data curation enables data sharing throughout the data-management life cycle to create new value for new user needs. One promising approach to data sharing in the cultural heritage domain is the metalevel ontology known as the CIDOCCRM (The International Committee for Documentation [CIDOC] Conceptual Reference Model [CRM]). The CIDOC-CRM is an object-oriented and extensible event-centred ontology for knowledge sharing in cultural heritage (CIDOC-CRM Special Interest Group, 2015). The CIDOC-CRM has been designated an international standard by the International Organization for Standardization (ISO 21127:2014). The CIDOC-CRM uses classes and property definitions in a formal structure to describe concepts and relationships. Because the CIDOCCRM is empirically derived, it is particularly well suited to intercultural information sharing. Cultural heritage applications of the CIDOC-CRM have demonstrated its usefulness in enhancing heterogeneous cultural heritage data in the form of a semantic framework. Data curation can consist of mapping data descriptions using the ontological structure of the CIDOC-CRM as an equalizing filter for data sharing. Records created in one repository can be shared among all repositories that utilize the CIDOC-CRM.

CIDOC-CRM mapping using structured data

Several studies have been conducted using the CIDOC-CRM ontologies to summarize structure and combine existing data in a generic ontological model and to map and formalize existing knowledge for interoperability with other data sets. Semantic mapping of cultural heritage between Encoded Archival Description (EAD) and the CIDOC-CRM ontology also has been studied (Bountouri and Gergatsoulis 2011; Gergatsoulis et al. 2010). Koutsomitropoulos, Solomou, and Papatheodorou (2009) conducted a case study to implement a prototype with Dublin Core metadata in digital object collections. Lin, Hong, and Doerr (2008) studied inference platforms by using OWL for mapping through the CIDOC-CRM for cultural heritage digital libraries. Stasinopoulou et al. (2007) investigated ontology-based metadata using the CIDOC-CRM; Theodoridou et al. (2010) studied CIDOC-CRM digital (CIDOC-CRMdig) in Resource Description Framework Schema (RDF/S) for modelling and querying provenance. Doerr and Iorizzo (2008) discussed how heterogeneous cultural heritage information could be enhanced using the CIDOC-CRM due to its flexibility for integration, mediation, and interchange. They presented a case study using core ontologies of the CIDOC-CRM. This study demonstrated the difficulties of linking documents and knowledge. For that reason, semi-automatic co-reference detection and correction were needed to match between identifiers and resources, because direct data input into a semantic network revealed ineffectiveness.

Ontological data sharing using open government data

Previous studies regarding open government data sharing using ontologies have focused on structured data sets rather than unstructured data sets in heterogeneous formats. Despite the increasing importance of ontologies in the Semantic Web, open government data -sharing through ontologies for data sharing has not been studied. A few published studies have applied ontologies as a means of enhancing data sharing using open government data (Fragkou, Galiotou, and Matsakas 2014; Park and Smiraglia 2014). Ontologies such as the Greek E-GIF ontology (eGovernment Knowledge Interoperability Ontology) and the CIDOC-CRM were studied using open government data and the nature of each ontology for data sharing was discussed (Fragkou, Galiotou, and Matsakas 2014). Park and Smiraglia (2014) found that current CIDOC-CRM ontologies support addresses only for Western address systems (i.e., assigning addresses based on street-house number system) rather than East Asian address systems (i.e., assigning addresses chronologically rather than geographically). Knowledge organization systems (KOSs), such as ontologies, can be used as tools for increasing access to and dissemination of government information to wider audiences for the public good (Hodge 2014). Ontological mapping of open government data records into the CIDOC-CRM can facilitate wider access because the CRM's empirically derived, event-based ontology is culturally neutral. Moreover, the CRM's compatibility with the RDF, a backbone of the Semantic Web, can help to generate greater access to open government data through the Semantic Web. The case study reported here is a demonstration of the power of the CRM to contribute to the general public through the network environment, freeing open government data from the silos of its naturally heterogeneous formats.

Methods

We used a mixed methods approach, in which we used quantitative co-word analysis of a population of open government data to empirically determine characteristics of the data records. The overall design of the study is represented visually in figure 1. Co-word analysis often is used as a domain analytical technique in knowledge organization research because it can be used with a large body of textual data to extract a core ontology for a domain. Domain analysis in knowledge organization is formulated as a methodological paradigm for discovering knowledge bases and revealing shared ontologies (Smiraglia 2015a; the methodology is described fully in chapter 6). Thus, our first step was to subject a population of Korean open government data to co-word analysis to improve our understanding of the knowledge base represented by the data elements. This analysis reveals key terms and phrases that co-occur and that recur with high frequency in the data set. Aside from the informative value of the co-word analysis, this is a very pragmatic approach to undertaking CIDOC-CRM mapping by associating key concepts with classes and properties. We followed this with a qualitative analysis of five carefully selected records, which we mapped using the CIDOC-CRM. A final step in the case study was

Figure 1.

Process of research design

to use FRBRoo, an extension of the CIDOC-CRM, to map the instantiation of data records in a typical data-sharing scenario.

Data collection was conducted from the official Korean Open Government Website (http://data.go.kr/) in two phases. In the first phase of data collection, all records in three cultural heritage categories—libraries, archives, and museums—were downloaded. All records were translated from Korean to English by the researchers before the co-word analysis. Each file or data set in any heterogeneous format was opened to identify representative cases for the qualitative phase by analyzing attributes of distinctive records in each file or data set, again in consultation with the results of the co-word analysis.

Results

Quantitative co-word analysis of titles

Co-word analysis involves the use of software for extracting keywords, terms, and phrases from a body of text for subsequent analysis. The terms extracted can be displayed in frequency distributions that help separate the core of the knowledge base—those most frequently occurring—from the granularity, made up of the traditional "long tail" of a Bradford-type distribution. (This methodology is fully described in (https://provalisresearch.com/) by first entering all of the translated records into QDA Miner. Then, using the WordStat module, we generated frequency distributions of keywords and frequently occurring phrases. We also used the KWIC (keyword-in-context) function to cross-check the occurrence of multiword terms with the phrase lists. In this manner, we were able to generate basic taxonomies, which can in turn be used to generate visualizations using multidimensional scaling. Goodness of fit requires low stress and high R² to develop a visualization that matches the actual term co-occurrence in the knowledge base well. Removing single-occurrence clusters sequentially improves goodness of fit.

We analyzed each segment—archives, libraries, and museums—separately and then in combination to derive diverse views of the data set. The archives data set was the smallest, with only seven records containing only 26 keywords, of which only 3 occurred more than once: archives, digital, and list. No multiword phrases were identified. Using the KWIC feature, we learned that the only recurring term was "digital archives." The libraries data set was much larger, with 42 records containing 3,840 keywords, of which 428 were unique. The phrase-finder revealed 3,375 phrases, of which 116 occurred 3 or more times, and 24 occurred 14 or more times (or more than 2.5% of the total). These were used to generate the visualization shown in figure 2 (stress ¼ 0.22536; R² ¼ 0.9250).

The visualization thus contains only the most prominent categories in the knowledge base, but reveals the diverse emphases in the data set. For example, library contents are important especially with regard to book holdings and conditions and new book arrivals. Continuing education programs are prominent. Both public libraries and digital libraries are central but distinct. Not shown here is the prominence of library information relevant to geography (e.g., province, city, and district), and library size and type in the Republic of Korea. To help provide context, WordStat has a keyword-in-context (KWIC) feature, which can be used to see how the core keywords have been used. For example, "book," the keyword with the highest frequency, is used predominantly in phrases such as "book database [list, information] in [place]" and in conservation statements such as "condition of book at [place]." From this simple analysis, we can see that the next several keywords combine to form a cluster about "book [information or condition] in [place]." This suggests that the CRM classes that will be most applicable are those identifying institutions (place appellation, legal body) and holdings (manmade object, information object) and various actors.

Figure 2.

MDS visualization of core terms from the libraries data set

Similarly, the museum data set was analyzed. In this set there were 1,543 keywords, of which 249 were unique. There were 989 multiword phrases, of which 42 occurred three or more times and 16 occurred five or more times (or more than 2.4% of the total). These were used to generate the visualization shown in figure 3 (stress ¼ 0.14649; R² ¼ 0.9636).

The visualization shows only one cluster, suggesting homogeneity in the data set, and that is dominated by the word "museum." The KWIC feature shows a predominance of statements of the form "[place name] Museum [gallery, information]" or "[place name, city, province, or district]." A small cluster of terms including "cultural" combines that word with "facility," "assets," "heritage," or "space." In addition to the CRM classes identified for libraries, there also is a need for "curation activity," "time span," and many simple "strings" giving actual information concerning the institutions, holdings, and activities.

Figure 3.

MDS visualization of core terms from the museums data set

Qualitative case study

Five cases were selected from the data downloaded from the Korean Open Government website for qualitative case study. Five cases were selected from the archives, libraries, and museums data sets, with care in the latter two instances to select both examples of management data and holdings data. The five cases are (1) archives of digital advertisement; (2) museum facility; (3) museum artifacts; (4) library of the Korea Research Institute of Bioscience and Biotechnology—booklist (as of March 2014); and (5) library management (August 30, 2015). See figure 4.

The cases then were mapped using the CIDOC-CRM, version 6.2. Mappings were based on those reported by Park and Smiraglia (2014), which were verified by key experienced CRM trainers. These are shown in figures 5–9.

Figure 5 displays the CRM mapping of "Archives of Digital Advertisement," a record from an archive of advertising collections in Korea. The record includes two cartoon advertisements. The first is titled "ChoonHyang and DoRyeong Lee," which is associated with a well-known Korean love story and folk tale that might be compared to the story of Romeo and Juliet in western culture. Just as "Romeo and Juliet" is associated with many iterations ranging from Shakespeare's original play to book titles, musical works, classical characters, and the like, the story of "ChoonHyang and DoRyeong Lee" here is used as the basis of a cartoon advertisement for LakHeui Chemicals. In figure 5, "ChoonHyang and DoRyeong Lee (E35 Title)" is the title of the digital adverting collection. The creator of this digital advertising is cartoonist (E39 Actor) Dal Boo Moon (E82 Actor Appellation). The duration (E54 Dimension) of the digital advertisement is 180 minutes (E52 Time-Span; E58 Measurement Unit),

Figure 4.

Five open government records

Figure 5.

CIDOC-CRM map of "archives of digital advertisement"

Figure 6.

CIDOC-CRM map of "museum facility"

and this digital advertisement was produced (E1 Production) in 1956 (E52 Time-Span). The advertiser (E74 Group) is LakHeui Chemicals (E40 Legal Body). The Archives of Digital Advertisement specify this advertisement as "group and corporate advertising (E55 Type)," "group PR (E55 Type)," and "group advertisement (E55 Type)" for LakHeui Chemicals (E40 Legal Body). The original is accompanied by an instantiation for overseas release.

Figure 6 displays the CRM mapping of "museum facility," a record for use by both museum facility administrators and museum users. For example, the Museum of Education (E39 Actor) opens from 9:00 to 18:00 and closes on Mondays, New Year, ChooSeok, and the day following ("next day"), national holidays (E62 String). The dimension of the museum (E54 Dimension) give a floor area (E55 Type) of 672.94 square metres (E60 Number; E58 Measurement Unit). Museum users need to pay for admission (E62 String). The Museum of Education provides some activities (E7 Activity) to the public, which take place at the facility with the address "Sosa-ro 482, Wonmigu (Chunuidong)" (E44 Place Appellation). The Museum also has events such as "exhibitions: education in the traditional era (E5 Event)," "exhibitions: education supplied (E5 Event)," "exhibitions: classrooms in the 60s and the 70s (E5 Event)," "exhibitions: students' life at school (E5 Event)," and "exhibitions: reference room (E5 Event)." The "Museum of Bow" is the Bucheon Bow Museum in the same location, with similar facility information. In both cases, the record is mapped using the CRM mapping for a "plan" (E101 Has intention to apply).

Figure 7 displays the CRM mapping of "museum artifacts," a record from one of Korea's national museums. This record provides specific information about specific museum artifacts: artifact name, dimensions, materials, provenance (i.e., "find spot"), physical details, purpose of the artifact, holding museum, and accession number. The first example is a banja, a metal Buddhist percussion instrument, donated by low-ranking local officials to a temple to wish for the long life and good health of the king. The second example is a floral design celadon, a celadon jar with floral medallion design in relief. The banja (E22 Man-Made Object) has identification number "chrysanthemum 000044-000" (E42 Identifier) and consists of a metal (E57 Material), specifically a copper alloy (E57 Material). The dimension of the banja is its diameter (E54 Dimension; E55 Type), 36.2 cm (E60 Number; E58 Measurement Unit), and its thickness (E55 Type), 7.4 cm (E60 Number; E58 Measurement Unit). The banja (E22 Man-Made Object) was created during the Goryo dynasty (E4 Period) and is a national cultural asset (E55 Type) of Korea (E74 Group). The national museum of Gyeongju (E87 Curation Activity), which is located in the North Gyeong-Sang Province (E53 Place), currently curates the banja. The banja was used for Buddhist religious ceremony (E17 Type Assignment) and represents the "study of ancient culture and heritage of the country" (E62 String). The banja is "is used to call the public in the Buddhist temple or to inform the urgent things" (E62 String). Fewer details are given about the floral design celadon, which is held by the same museum.

Figure 7.

CIDOC-CRM map of "museum artifacts"

Figure 8.

CIDOC-CRM map "Library of the Korea Research Institute of Bioscience and Biotechnology—booklist (as of March 2014)"

Figure 9.

CIDOC-CRM map of "library management"

Figure 8 displays the CRM mapping of the "Library of the Korea Research Institute of Bioscience and Biotechnology—booklist (as of March 2014)," a special library in Korea. The record contains details about a journal article. The record for the article (F5 Item) has a record identification number, 7006762 (E42 Identifier), and a library article identification number, KEI 2013–2 (E42 Identifier). The title (E35 Title) of this article is "Environmental Aspects and Management of Hypothetic Zones," the author (E39 Actor) is Hyeon Yoon Jeong, and its publisher (E40 Legal Body) is the Korea Environment Institute (KEI). The library call number for this article uses the Library of Congress classification symbol QP601 (E55 Type; general works on enzymes within animal biochemistry), a local title number M49, date 1955 (E44 Type), andv. 1 (E55 Type). Figure 9 displays the CRM mapping of "library management." The record is useful for managerial purposes because the data set includes library type, library hours, total holdings, total numbers of seats, organization that operates (or owns) the library, location, and construction year. For instance, the BaekHab Children's Library (E22 Man-Made Object), a small public library (E55 Type), was constructed (E12 Production) on August 31, 2015 (E63 Beginning of Existence) and is operated by Yong Sangu District (E40 Legal Body). The address (E44 Place Appellation) of this library is "19-2 Huam-ro 15-gil" (E45 Address), "Yongsan-gu District" (E45 Address), and "Seoul City" (E45 Address). The telephone number is 755–0525 (E51 Contact Point). Library hours are from 10:00 to 17:00 (E61 Time Primitive) and the library is closed on Sundays (E52 Time-Span). The total number of seats in the library is 8,863 (E54 Dimension; E60 Number) seats (E55 Type). Similar details are given for the Book Café CheongMaRu.

Discussion

CIDOC-CRM and FRBRoo ontologies in KO

The mappings shown above are simple and straightforward. CIDOC-CRM is an event-centred ontology for cultural heritage. In our data sets, books (as artifacts) are controlled according to their acquisition, holdings availability, or condition in a diverse but related set of government repositories. The data map quite well to the CRM, which tells us that scalability—extending the mapping to the entire set of repository data—is possible.

Of course, an interesting problem then emerges, which is how to control the instantiation of the data. Instantiation is a phenomenon of information objects that takes place when the same content has several carriers (Park and Smiraglia 2014). For example, the text of this article exists in manuscript in Microsoft Word, but also in the PDF format. If we map our data sets using CRM, we create an instantiated set of mapped data. How, then, can an information system disambiguate mapped data?

An extension of the CIDOC-CRM is the FRBRoo ontology (Functional Requirements for Bibliographic Records—object-oriented; http://www.cidoccrm.org/frbr/). The FRBR conceptual model is based on understanding that an

Table 1.

FBRRoo/CRM Mapping of Archives of Digital Advertisement

abstract "work" takes intellectual form in its "expressions," published form in its "manifestations," and physical form in "items" (which, of course, may be digital). These entities, sometimes represented as W-E-M-I, are FRBR's explication of instantiation in the library community. The object-oriented version of FRBR has been harmonized with the CRM, and therefore may be used to disambiguate mapped data such as those in this case study.

In our study, each record is a work, each version of it is an expression (so that we have the open government data expression and our CRM-mapped expression), and their occurrence in available data sets constitutes manifestation. In addition, several of the records identify objects that can be interpreted as works, and in one case textual works, which can be mapped to FRBRoo classes as a means of enriching the mapping. Tables 1–5 display parallel mapping of our five cases using FRBRoo as an extension of CRM mapping. In the tables, to the best of the capacity of the tabular format, we have attempted to draw parallels between the CRM and FRBRoo categories. For example, in several cases, E40 Legal Bodies (CRM) are the parallel mappings for bodies that also serve as F11 Corporate Bodies (FRBRoo). Similarly, where FRBRoo has inherited the CRM category, such as E35 Title, no parallel is shown.

Table 1 displays parallel mapping between FRBRoo and CIDOC-CRM for the record "Archives of Digital Advertisement." This mapping shows the alignment between FRBRoo and CIDOC-CRM, which is useful for extending the

Table 2.

FRBRoo/CRM Mapping of Museum Facility

two ontologies together for digital records of archival materials. Noticeable are points such as E35 Title where FRBRoo refers to CRM classes, and the overlap between F11 Corporate Body and F12 Nomen for the two corporations that are both producers and subjects of the cartoons, and F12 Nomen and E35 Title for the characters that are also subjects of the cartoons.

Table 2 displays parallel mapping between FRBRoo and CIDOC-CRM for the record "museum facility." In this case, the FRBRoo mapping provides an intersection between the appellations for the museums and their place of location.

Table 3 displays parallel mapping between FRBRoo and CIDOC-CRM for the record "museum artifacts." Once again, we see an overlap between the FRBRoo F12 Nomen and F35 Nomen Use Statements for the objects identified in the record as CRM E22 Man-Made Objects. This points out the dual usage wherein artifacts are represented both as objects and as names of kinds of things. Also, FRBRoo allows the characteristics of the objects themselves to be represented as F3 Manifestation Product Types. In this way the mapping is enriched by the use of the two ontologies together.

Table 3.

FRBRoo/CRM Mapping of Museum Artifacts

Table 4 displays parallel mapping between FRBRoo and CIDOC-CRM for the record "Library of the Korea Research Institute of Bioscience and Biotechnology (as of March 2014)." The specific objects identified in the record are texts available in the library; thus, FRBRoo is very useful for extending the mapping to the bibliographic characteristics of the texts.

Table 5 displays parallel mapping between FRBRoo and CRM for the record "library management (31 August 2015)." The record contains standard information for the administrative purposes of managing libraries in the Republic

Table 4.

FRBRoo/CRM Mapping of Library of the Korea Research Institute of Bioscience and Biotechnology—Booklist (as of March 2014)

of Korea. As was the case in table 2, the representation of "museum facility," there is little overlap between the two ontologies because this record contains little bibliographic information. However, F44 Bibliographic Agency supplies a potential point of connection should the libraries in this record also be represented elsewhere in the data set as producers of bibliographic data for their holdings.

Finally, figure 10 shows how we used FRBRoo to map each instantiated record. Each record in a data set is an information object (F73), which is expressed (F2) in the form of the tangible data in the data set. Researchers (E39 Actor) translated the original records from Korean to English, creating a second set of information objects, which are manifestation product types (F3) that also, at least in this case, are manifestation singletons (F4). Each has meta-data concerning the actors and the expression creators such as the original, the translated, and the CRM-mapped. The end object is inherent in a set of information carriers (E84) that requires disambiguation. We have added value by extending the mapping of the instantiated records with FRBRoo.

Conclusions

We applied a metalevel ontology for cultural heritage information sharing to a set of selected Korean open government data. Our analysis of the population of data sets from which the cases were drawn demonstrates the compatibility of these cultural heritage open data sets with the event-based CIDOC-CRM ontology. Further, our analysis extends the mapping by the use of FRBRoo to provide disambiguation for the mapped data. In this way, we have demonstrated one

Table 5.

FRBRoo/CRM Mapping of Library Management (August 31, 2015)

approach to the use of classical facet analytical theory with the CRM and FRBRoo ontologies by using CRM to represent event-based data and FRBRoo to add dimensionality in a manner similar to the use of auxiliary devices employed in many classification systems (Smiraglia 2017, 6). This study is original—to the best of our knowledge, FRBRoo mapping with CRM using open government data has not been discussed previously.

Further, we have borrowed methods and theoretical framing from knowledge organization. In knowledge organization, the concept is the basic element of any ontology, and it is the ordering of concepts that produces a useful knowledge organization system. CIDOC-CRM uses the concepts embedded in cultural heritage metadata, reordering them by event-based categories and properties. To this we have added the FRBRoo categories and properties that disambiguate instantiation sets. We have shown that knowledge organization domain analytical techniques can be applied to the analysis of potential data for ontological

Figure 10.

FRBRoo extension

mapping. We have demonstrated the utility of parallel use of the CIDOC-CRM and its FRBRoo extension as valuable ontologies for mapping digital cultural heritage data sets to facilitate information sharing. Also, this study contributes to the cross- or metainstitutional integration of curation across institutional boundaries in cultural heritage institutions such as libraries and museums as the imperative for cultural synergy and the role of information institutions (Park and Smiraglia 2014) by reusing open government data in heterogeneous formats.

Ontologies derived from open government data are culturally neutral because no biases are represented when a resource is described, meaning no biased judgments of culture. This study focuses on the internal records of the open government data, which themselves were generated from the repository. Ontological mappings from this study can be applied to both (1) the virtual catalogue with the data components and (2) the visualization of KOS, which can provide useful navigational maps.