Pampa Azul is a governmental initiative in Argentina that supports research activities through oceanographic campaigns and promotes interdisciplinary cooperation between institutes of marine research in areas of national jurisdiction. The Argentine Continental shelf houses commercial sheries, biodiversity, hydrocarbon basins and mineral deposits of great economic and ecologic importance. In particular, the San Jorge gulf6 has ecologic and geographic features that brought together broad and sustained oceanographic research activities. The gulf is located in the central region of the Patagonian ocean litoral, where oil and shing industries coexist along with tourism, which gives rise to ongoing and acute environmental risks. For this reason, accurate and frequent oceanic sampling and measurement is of critical social, economic and ecologic importance. Oceanographic campaigns funded by the Pampa Azul initiative are the basis of scienti c research at sea, yielding huge amounts of data, highly heterogeneous in types and formats, and scattered across distributed data repositories.
Oceanographic research and e cient management of the collected data often appear to be two widely separated worlds. Data managers consider the careful collection, management and dissemination of research data as essential for the e ective use, while researchers consider data management as a merely technical issue, of little relevance for their interests. Consequently, data management is often insu ciently planned, if at all, and receives very low priority and budget. As part of governmental policies towards ocean management, researchers from Argentine scienti c institutions are required to disseminate their activities and accomplishments to give greater visibility to the national e orts in this eld. 6 http://www.pampazul.gob.ar/areas-prioritarias/golfo-san-jorge/ For individual researchers, this situation presents a di cult challenge regarding discovery, access, and integration of data which they need to conduct scienti c inquiries. As speci c cases of these problems, we can mention (i) data level conicts caused by di erences that arise in data domains due to multiple possible representations and similar data interpretations, and (ii) lack of an integrated knowledge infrastructure which hinders the ability of researchers to analyze potential discovery scenarios if more than one repository is involved. For example, they may be interested in knowing if in a marine region there are CTD data available7, provided by oceanographic campaigns in the region of interest performed by research vessels not belonging to Pampa Azul.
The Semantic Web (SW) [
In this short paper we present the initial steps in the creation of a
oceanographic linked dataset using information from the oceanographic campaigns
of Pampa Azul. To achieve consistency, discoverability and make the
datasets readable by machines and humans, we use di erent controlled vocabularies
among them NERC Vocabulary Server8 together with the geospatial standard
for the semantic web GeoSPARQL [
The publication of the dataset involves di erent steps that are described below and the architecture of such a process can be consulted in10.
{ Imput data: National Marine Data System (NMDS11) is a web platform that allows publishing datasets of oceanographic campaigns that was sampled in the Argentine sea. These datasets are composed of (i) metadata of the oceanographic campaigns (name of the campaign, vessel, dates, people and institutions involved, geographical coverage among others.) in XML format, and (ii) data recorded by the vessel in its trajectory, which contains the information of the measured variables (pressure, salinity, temperature, depth, 7 CTD: instrument used to measure the conductivity, temperature, and pressure of seawater. 8 https://www.bodc.ac.uk/resources/vocabularies/vocabulary_search/P01/ 9 http://data.rvdata.us/ 10 https://github.com/cenpat/pa-lod/blob/master/images/framework-PA.png accessed at April 2018 11 http://www.datosdelmar.mincyt.gob.ar/index.php positions where the variable was sampled among others) and additionally contains information of the equipment that was used to sample (CTD, Termosalinometer, etc.). { Data extraction and cleaning: Metadata and data of campaigns are manually extracted from the NMDS repository and their content are processed using OpenRe ne tool12. There, the columns are cleaned and converted to standardised data types such as dates, numerical values, etc. and empty columns are removed. { URI strategy: Currently URIs for the resources belonging to oceanographic campaigns follow the pattern:
http://data.pa.gob.ar/lod/ftypeg/fconceptg/fIDg The domain only be used for the publication of pampa azul information and not include the name of any organization, as they may evolve over time. ftypeg can take any of the following values: resource for the HTTP URI of a resource, and page and data for that resource's HTML and RDF documents respectively. fconceptg gives a hint as to what this resource is about by referring to the class to which that resource belongs, for example, Cruise, Dataset, Person etc. fIDg for the unique identi ers we use the ones provided in the original datasets, normally identi ed with a Universally Unique Identi er (UUID). { Conversion to RDF: Data are converted to RDF triples using RDF Rene13 that allows users to go through a graphical interface describing the RDF scheme skeleton which speci es the subject, predicate and the object of the triples to be generated. The next step in the process is to set up pre xes. Since datasets include localities, locations and research institutes, we set up pre xes for well-known vocabularies such as FOAF, Dublin Core, NERC parameter codes and GeoSPARQL. To see the resulting graph after the conversion see the link14. { RDF storage: The transformed data have been published, and can to be visualised through GraphDB which is a highly e cient and robust graph database with RDF and GeoSPARQL support. It allows users to explore the hierarchy of RDF classes, relationships among these classes, etc. 3
Use Case: Complementing Pampa Azul Information With R2R The following use case explores the R2R oceanographic linked dataset using a federated query to retrieve all the trajectories that exist in the R2R dataset and that are within a polygon de ned in the FILTER clause, this polygon de nes the exclusive Argentine economic area, so this query is interesting since it allows to know which cruises traveled that region at some time. 12 http://openrefine.org/ 13 http://refine.deri.ie/ 14 https://github.com/cenpat/pa-lod/blob/master/images/pa-graph.png cessed at April 2018 acPREFIX geosparql : <http :// www . opengis . net / ont / geosparql #> PREFIX geof : <http :// www . opengis . net / def / geosparql / function /> PREFIX sf : <http :// www . opengis . net / ont / sf #> SELECT ? track WHERE { SERVICE <http :// data . rvdata . us / sparql > { ? track a sf : LineString . ? track geosparql : asWKT ? gWKT FILTER ( geof : sfWithin (? gWKT , " POLYGON (( -63.80859375 -41.45713698292349 , -47.109375 -41.457136982923494 , -47.109375 -50.9151558824997 , -63.80859375 -50.9151558824997 , -63.80859375 -41.457136982923494)) " ^^ sf : WktLiteral )) } } 4
In this short paper we presented an overview of our initial e orts to create a Linked Oceanographic Dataset, reusing Ontological Design Patterns and speci c vocabularies of this domain. In order to test our dataset we extracted metadata in XML format from NMDS. In this initial stage (April 2018) our platform stored 618K RDF triples with a total of ten classes instantiated. Also for the user to exploit the dataset we de ne SPARQL queries that can be accessed through the link15. Finally, the user can visually explore the dataset, accessing the following link to the GraphDB interface16 (user: pauser password: pauser). We have shown that RDF can be used to represent metadata about oceanographic campaigns of Pampa Azul initiative in a useful way, but there is still a lot of fruitful work to be done. Particularly as future work, we need to develop information visualization interfaces that allow non-expert users to explore the data. For this we have explored map4rdf17 a mapping and faceted browsing tool for exploring and visualizing RDF datasets enhanced with geometrical information.