Title: Cobis Linked Open Data Portal
Team: Synapta (& Friends!)
The CoBiS Linked Open Data pilot project provides a unique point of access to the collections of seven of the 65 specialized libraries forming the CoBiS network. We merge cataloguing data from five different pieces of software, also linking to multimedia content. The interlinking focuses on Wikidata and VIAF, and the reference ontologies are BIBFRAME and Schema.org. During the implementation of the project, we developed and released as open source JARQL, an innovative triplification software (https://github.com/linked-solutions/jarql), and we developed OLAF (https://olaf.synapta.io/), a mix’n’match-style interface to crowdsource the interlinking of authority files.
The CoBiS (i.e., “Coordinamento delle Biblioteche Speciali e Specialistiche di Torino”) is an informal network of 65 specialized libraries, collaborating to provide continuing professional development and to offer a better service to their users. CoBiS libraries are heterogeneous from any point of view: they have diverse collections, use different cataloging software (Clavis, ErasmoNet, BIBLIOWin, SBNWeb, and Sebina OpenLibrary), and various online public access catalogs (if available).
# CoBiS Linked Open Data Portal =
The CoBiS Linked Open Data Project provides a unique point of access to the collections of CoBiS libraries. We don’t just make CoBiS bibliographic data more interoperable among them. We also open them to facilitate collaboration with online communities and connect them to the Linked Open Data Cloud.
In our submission, we present the new CoBiS Linked Open Data portal. Through this semantic gateway, you can explore data using a search engine, or navigating different sections about authors, books, and data journeys. Our video shows, for instance, Galileo Galilei’s page, dynamically generated through SPARQL queries.
In Galilei’s page you have, on the left, biographical information and a list of interlinked resources coming from Wikidata and other bibliographic repositories. On the right, you see an infobox with Galilei’s data and an image, both coming from Wikipedia, thanks to the leverage of Linked Data. Moreover, clicking on the RDF button, all the triples of Galilei resource can be directly fetched. Finally, at the bottom of the page, all Galilei’s books inside the Cobis database are shown. To explore information on such books, you can click on one of these boxes or you can use the search bar, looking for something not listed.
Our video shows an example search for the Sidereus Nuncius.
On the left side of the Sidereus Nuncius page, you see bibliographic details with a collection of interlinked resources. Exploiting the power of Linked Data, we are able to read the Internet Archive digital copy of the book. The physical copy is currently available in some CoBiS libraries, with more details about the options to actually access it.
# CoBiS LOD Project Architecture: the Ontologies
We now focus on the Linked Data stack. We used two main ontologies: Bibframe and Schema.org. We also used selected properties from RDFS, OWL, DCTerms, FAOF, and Culturalis (http://culturalis.org/ – used to describe CoBiS libraries as cultural heritage institutions).
# CoBiS LOD Project Architecture: Triplification with JARQL
As far as the project architecture is concerned, you can get an overall picture at the following URL: https://demo.cobis.to.it/wp-content/uploads/2017/04/architettura-cobis.png.
We started from heterogeneous sources, with the aim of combining them into RDF data. To reach this purpose, we used both RML (in particular in a first phase of the project) and our newly developed JARQL library, that turns JSON into RDF using a mapping written as a SPARQL construct query. In the video, we show how JARQL works, taking the example of a JSON file containing information about the Sidereus Nuncius. Using a SPARQL query, we describe the transformation, obtaining an RDF file as result. (You may try JARQL live at http://jarql.synapta.io/.)
# CoBiS LOD Project Architecture: Crowdsourced Interlinking with OLAF
During the graph building, we also enriched data with external sources. We used both automatic algorithms and manual approaches. For the latter, we exploit OLAF (Open Linked Authority File), our crowd-sourcing interface for creating an authority file. For instance, in the video, we want to match the correct Wikidata entity to a person named “Galilei”, who wrote the books mentioned on the right side. Calling Wikidata SPARQL API, we obtain different entities that you can choose to identify the exact matching. In this case Galileo Galilei.
(You may try OLAF live at https://olaf.synapta.io/ selecting “CONNETTI” near “Autori dei libri del CoBiS” and registering, in order to annotate the suggested triples with your identity.)
All the interlinks, discovered via OLAF and the other enriching techniques, make the portal able to access to external resources: VIAF, Wikidata, the Italian National Catalog, Treccani.
# CoBiS LOD Project Architecture: Full Stack Linked Open Data
Finally, as result, we have exposed data through a SPARQL endpoint that feeds the Cobis portal. We want to stress this aspect: in a few weeks, the CoBiS LOD Project portal will be online with its full Linked Data stack, including a public SPARQL end-point (configured to support federated queries), a full dump of the RDF data, LodView to dereference URIs, etc.
# What’s Next
As next steps, we want to enrich the interlinking of our data with the Linked Open Data Cloud. In particular, websites as WikiSource (and the aforementioned Internet Archive) have many contact points with our project (and, e.g., Wikidata allows us to locate the Wikisource URL for a given work).
Moreover, we are going to involve other members of the Cobis in the Linked Open Data project (7 are in, 58 to go!), evolving the portal towards a full fledged Linked Data digital library.
The CoBiS Linked Open Data portal is going to be available by May 2017 at http://dati.cobis.to.it/.