Generating Conference Linked Open Data in One Click

Abstract

In this paper we describe cLODg2 (conference Linked Open Data generator - version 2), a tool to collect, re ne and produce Linked Data about scienti c conferences with their associated publications, participants and events. Conference metadata collected from di erent unstructured and semi-structured resources must be expressed with appropriate vocabularies to be exposed as Linked Data. cLODg2 facilitates this task by providing a one-click work ow to generate data which is ready to be integrated in the ScholarlyData.org dataset. cLODg2 is an open source project, which has the aim to foster the publication of scholarly Linked Open Data and encourage collaborative e orts in this direction between researchers and publishers.

-

Scholarlydata [ 4 ] is the evolution of the Semantic Web Dog Food (SWDF) dataset3. The SWDF corpus was the rst considerable e ort to o er comprehensive semantic descriptions of conference events [ 3 ], collecting linked data about papers, people, organizations, and events related to academic conferences.

A comprehensive description of Scholarlydata can be found in [ 4 ], while in this paper we provide technical details about cLODg2, the Open Source tool4 that supports data generation for Scholarlydata. cLODg2 (conference Linked Open Data generator - version 2) provides a one click process for the conference metadata publication work ow. cLODg2 has been used to refactor the SWDF dataset and to gather and publish new conference metadata5. The tool provides an easy process to generate Linked Data which can be directly added to the ScholarlyData dataset. 3 SWDF: http://data.semanticweb.org 4 https://github.com/anuzzolese/cLODg2 5 Amongst other it has been used for ESWC conference since 2014 http://2016.

eswc-conferences.org

cLODg2 - publishing Conference Semantic Data The main goal of cLODg2 is to facilitate the generation of conference Linked Data which can be readily integrated in the Scholarlydata6 dataset. Scholarlydata [ 4 ] is the evolution of the SWDF dataset [ 3 ] based on an improvement of the Semantic Web Conference (SWC) Ontology7, the Conference Ontology8[ 5 ], which improves SWC adopting best ontology design practices. The necessary steps to add conference data to Scholarlydata are: (i) Data acquisition, (ii) Linked Data generation, (iii) Linked Data enrichment and (iv) Linked Data Publication.

The Data acquisition step, to be done by the user, consists of acquiring metadata about the conference, generally exported from a conference management system. We currently support data acquisition from CSV les9. Additionally, Linked Data represented with the SWC ontology can be used as initial input10.

Starting from provided input cLODg2 performs two sequential steps: Linked Data generation and data enrichment. Figure 1 shows the system architecture, including all accessed services and technologies, modelled as an UML activity diagram. The initialisation step merely consists of con guring a property le to point to (i) the collected CSV les containing the input data and (ii) the D2RQ mapping that will serve for converting CSV les to RDF. A D2RQ mapping for dealing with easychair data is provided by default, but expert users can change this to import ad hoc CSV les.

The Linked Data generation activity is composed of the following steps: { Data gathering. This action merely represents the system fetching data from the speci ed location. We remark that for the sake of simplicity we x the easychair model for the input data, but that this can be easily con gured for multiple data gathering support. { RDB population. This action aims at populating a relational database (RDB) from the CSV les gathered from the previous action. The RDB is based on HyperSQL (HSQLDB)11, which is a lightweight open-source Java database; { D2R conversion. The previous action, i.e., RDB population, is preparatory to this step. In fact, cLODg2 relies on the D2R framework [ 1 ] to perform the conversion of a non-RDF source to RDF. The conversion is guided by the mapping provided as input. This mapping is described by using the D2RQ 6 http://w3id.org/scholarlydata 7 http://data.semanticweb.org/ns/swc/swc_2009-05-09.html 8 Refer to http://w3id.org/scholarlydata/ontology/conference-ontology.owl to obtain the OWL source code and to http://goo.gl/4lOHSk to obtain the HTML documentation of the Conference Ontology. 9 A simpli ed example of such data, exported from easychair.org can be found at https://github.com/anuzzolese/cLODg2/tree/master/csv_samples 10 Example dump at https://github.com/AnLiGentile/cLODg/tree/master/ resources/swdf_samples 11 http://hsqldb.org mapping language [ 2 ]. cLODg2 is released along with a default mapping for easychair data and targets two distinct alternative datasets: the SWDF and Scholarlydata.

The Linked Data enrichment activity is composed of the following actions: { Reasoning-based alignment. Input of this action are the RDF triples produced by the Linked Data generation activity. The output is the materialisation of a set of RDF triples that enable the alignment to other ontologies and vocabularies, i.e., the SWDF ontology, SPAR12, Dolce D013, the Organization Ontology14, FOAF, SKOS, icatzd , and the Collections Ontology15. The alignment triples are materialised by means of OWL-DL reasoning, which is enabled by the Apache Jena inference layer; { Linking to other Linked Datasets. This action is aimed at producing instance level alignments, expressed via owl:sameAs axioms. The target linked datasets are ORCID16 and DOI17. ORCID provides persistent digital identiers for scienti c researchers and academic authors. A digital object identi er 12 http://www.sparontologies.net 13 http://www.ontologydesignpatterns.org/ont/dul/d0.owl 14 https://www.w3.org/TR/vocab-org 15 http://purl.org/co 16 http://orcid.org 17 https://www.doi.org (DOI) is a serial code used to uniquely identify digital objects, particularly used for electronic documents. The alignments to ORCID are produced by relying on the public API provided by ORCID18. The references to DOI are produced by relying on the API provided by Crossref19, performing a search on each article title.

The Linked Data Publication step, which is the last action in the cLODg2 workow, has to be done by the user and consists of submitting produced data to Scholarlydata.org. 3

Conclusions This paper describes cLODg2, a tool to collect, re ne and produce Linked Data to describe scienti c conferences and their publications, participants and events. The main contribution of this work is an open source tool to support the production of metadata for conferences and scholarly data which is ready to be integrate in the ScholarlyData dataset, with minimal user e ort. Future work will be mainly focused at addressing data quality and reduce duplications and misspelling in the data. 18 http://members.orcid.org/api/introduction-orcid-public-api 19 http://www.crossref.org/guestquery

References

1. C. Bizer and R. Cyganiak. D2R Server - Publishing Relational Databases on the Semantic Web . In Proc. of ISWC2006 Poster&Demo , 2006 .

2. C. Bizer and A. Seaborne. D2RQ - Treating Non-RDF Databases as Virtual RDF Graphs . In Proc. of ISWC2004 posters , 2004 .

3. K. Moller, T. Heath, S. Handschuh , and J. Domingue . Recipes for semantic web dog food: The eswc and iswc metadata projects . In Proc. of ISWC'07/ASWC'07 , pages 802 { 815 , Berlin, Heidelberg, 2007 . Springer-Verlag.

4. A. G. Nuzzolese , A. L. Gentile , V. Presutti , and A. Gangemi . Conference Linked Data Our Web Dog Food has gone gourmet . In Proc. of ISWC2016 Resource Track , page to appear, 2016 .

5. A. G. Nuzzolese , A. L. Gentile , V. Presutti , and A. Gangemi . Semantic web conference ontology - a refactoring solution . In The Semantic Web: ESWC 2016 Satellite Events , page to appear. Springer, 2016 .