SPARQL query example 

Context in source publication

Context 1

... with the number of triples involved in each case. Finally, although the MS schema defines „ language ‟ information (both names and codes) as simple xs:string 20 type elements, we used the FAO ontology for language codes (ISO638-3). Currently, our dataset includes 298 triples with language code information, using 19 different codes. Besides using existing vocabularies, we defined links between our instances and external instances whenever possible. We used the sameAs and seeAlso properties to point to the relevant records in the DBLP, the Virtual International Authority File (VIAF) and the LOC 21 Authorities Names . Table 11 lists the correspondences between agents and documents with external data. Data categories are, by large, the most prolific elements in the MS schema. As we have already mentioned, the schema includes up to 807 different enumerations that were translated into instances of relevant classes in the final model (see the example in Figure 5 above). We undertook the task of linking these instances to relevant ISOcat data categories and to DBpedia whenever possible. In Table 12 we list the data categories used: the first column indicates the class; the second column shows the number of instances for each class; and the third and fourth columns show the number of linked instances to ISOcat and DBpedia respectively (we defined a ms:dcr property to link to ISOcat and used sameAs property with DBpedia). As corresponding explained in tools the ”. first Services section (and of this tools) paper, are resources we come from which a are scenario distinguished where corpus from and other lexicons resources are described in their in resourceComponentType the META-SHARE node . and stored in our institutional e-Repository, Due to these disparate whereas descriptions services are it registered is not possible in the to BioCatalogue. directly map the In BioCatalogue, information in a the service input is SoapServices defined as “ a container descriptions for into ServiceDeployments the target MS toolService and variants component. ” , where a variant This implies is either that a we SoapService need to or merge RestService the MS , and and a ServiceDeployment BioCatalogue ontologies. is a particular The solution running is instance to map of the the service variants . Services of a given also BioCatalogue include information service as about different testing. MS On resources the other , i.e. hand, as subtypes in the MS of model, the MS a Service service . is Table 13 defined as shows a “ form the resulting in which merged NLP tasks model. are realized and delivered corresponding tools ”. Services (and tools) are resources which are distinguished from other resources in their resourceComponentType . Due to these disparate descriptions it is not possible to directly map the information in the input SoapServices descriptions into the target MS toolService component. This implies that we need to merge the MS and BioCatalogue ontologies. The solution is to map the variants of a given BioCatalogue service as different MS resources , i.e. as subtypes of the MS Service . Table 13 shows the resulting merged model. LOD is the perfect place to merge and combine data. As explained in the preceding section, by reusing data from external vocabularies we were able to re-define some MS components such as Person and Organization in a much more standard way (via FOAF). Similarly, for Documents, the BIBO ontology allowed us to merge our bibliographical database into the dataset. In the original MS schema, documents could only be attached to resources, either as the „documentation‟ of the whole resource or as „reports‟ relevant for its validation, usage, annotation and evaluation aspects. In the converted model, documents include articles, manuals, reports, videos, etc. and constitute an important element of the dataset. We assume that everything can be documented either directly (by means of the ms:documentation or ms:validationReport properties among others) or indirectly (by means of dc:subject or dcterms:references properties). Thus, any class and instance in the dataset may be linked to some relevant document . For example, the class SOAPservice may be referenced by some video or article. Similarly, a Task instance such as namedEntityRecognition or a Standard instance such as LMF may be also documented. We use the dc:subject and dcterms:references properties to link documents to relevant parts of the data set. Integrating disparate, yet related, datasets into a single repository has obvious benefits, especially, to the maintainability, interoperability and integrity of the data. All of this has positive consequences both on cost and functionality. We also get an additional advantage from the properties of the RDF/OWL/SPARQL framework, which makes data exploitation simple and efficient. For example, if our user wants to know about Named Entity Recognition (NER), we can get all relevant data with a very simple query (Figure 6). With this simple query we easily retrieve „everything that has to do with NER‟. In this case we get Articles , Reports and Projects dealing with NER as well as Services performing such a task. In addition, the seeAlso property suggests us to check namedEntity . All data is stored at the SPARQL endpoint: We run a data browser located at The benefits of our exercise can be summarized as follows. On the one hand, the final RDF/OWL model is much simpler than the original XSD schema: it clearly differentiates between Classes and Properties and avoids problems typical of XML syntax, such as semantic ambiguity and order constraint. This is essential for mapping purposes. On the other hand, the open world assumption of RDF/OWL allows to naturally integrate objects from different schemas and to add further extensions, making merging of different models straightforward. We were effectively able to merge Service and Document ontologies into the MS model in a natural way. Moreover, linking data has an additional benefit: The resulting model not only includes linking between merged catalogues but also linking to external data. Last but not least, the RDF/OWL version of the registry (loaded in a SPARQL end point) not only provides a single unified repository but also facilitates the exploitation of the metadata records by the end-user. The success of our experiment encourages us to apply it on a larger scenario, such as the CLARIN Component Registry. The idea behind the CMDI approach was that identifying components blocks in the original schemas would improve interoperability among models. However, the proliferation of components in the Component Registry eventually becomes a critical problem. Conversion to RDF/OWL paves the way for more ambitious goals such as being able to derive a general ontology accounting for the different underlying schemas. The work reported has been co-funded by the "Fons europeu de desenvolupament regional (FEDER), Programa operatiu FEDER de Catalunya 2007-2013, Objective 1". Broeder, Daan, Oliver Schonefeld, Thorsten Trippel, Dieter Van Uytvanck and Andreas Witt. “A pragmatic approach to XML interoperability – the Component Metadata Inf rastructure (CMDI).” Presented at Balisage: The Markup Conference 2011, Montréal, Canada, August 2 - 5, 2011. In Proceedings of Markup Technologies, vol. 7 (2011). doi:10.4242/BalisageVol7. Haslhofer, Bernhard and Klas, Wolfgang. (2010) A survey of techniques for achieving metadata interoperability. ACM Comput. Surv . 42, 2, Article 7 (March 2010), 37 pages Hunter, Jane and Lagoze, Carl (2001) Combining RDF and XML schemas to enhance interoperability between metadata application profiles. In Proceedings of the (WWW '01). ACM, New York, NY, USA, 457-466. Lam, H. Y., Marenco, L., Shepherd, G. M., Miller, P. L., Cheung K. H. (2006) Using web ontology language to integrate heterogeneous databases in the neurosciences. AMIA ... Annual Symposium proceedings / AMIA Symposium . AMIA Symposium, pp. 464-468. Tsinaraki, Chrisa and Christodoulakis, Stavros (2007) Interoperability of XML schema applications with OWL domain knowledge and semantic web tools. In IS - Volume Part I (OTM'07), Robert Meersman and Zahir Tari (Eds.), Vol. Part I. Springer-Verlag, Berlin, Heidelberg, 850-869. Windhouwer, M., & Wright, S. E. (2012). Linking to linguistic data categories in ISOcat. In C. Chiarcos, S. Nordhoff, & S. Hellmann (Eds .), Linked data in data and language metadata (pp. 99-107). Berlin: Springer. Xiaoshu Wang, Robert Gorlitsky, Jonas S Almeida (2005). From XML to RDF: how semantic web technologies will change the design of 'omic' standards. In Nature Technology, Vol 23, No 9, pp 1099-1103, Sep ...