Section: New Results

Linked Data Access

Participants : Serena Villata, Luca Costabello, Fabien Gandon.

We designed and developed a context-aware access control framework for the Web of Data called Shi3ld (http://wimmics.inria.fr/projects/shi3ld/ ). The framework protects access to SPARQL endpoints, and it adopts Semantic Web languages only, as in the philosophy of the Web of Data. The innovative feature of the proposed framework consists in evaluating the accessibility to the data considering the attributes of the users. These attributes are defined following three main dimensions: user, device, and environment. The evaluation of the model shows that access control comes with a cost but it guarantees the protection of the data published on the Web of Data. The results of this research activity have been published in international conferences in the area of Artificial Intelligence (ECAI, [35] ) and the Web (WWW [61] , HT [34] ).

On the same line, we have proposed a framework for attaching the licenses to the data resulting from a query on the Web of Data. The rationale is that the licenses associated to the data returned by the query are selected, and using a number of rules their compatibility is assessed. If the licenses are evaluated compatible, then they are composed into a composite license which is released to the user together with the data. The results of this research have been published in the COLD international workshop [74] and in the ISWC international conference [75] (best poster award). These two research lines have been performed in the context of the DataLift ANR project.

The PhD thesis of Luca Costabello, directed by F. Gandon and I. Herman (CWI and Semantic Web Activity Lead at W3C) investigates Web of Data interaction from mobile environments. Two main research activities have been carried out in 2012: i) PRISSMA (http://wimmics.inria.fr/projects/prissma ), an adaptive rendering engine for RDF and ii) Shi3ld(http://wimmics.inria.fr/projects/shi3ld ), a context-aware access control framework for Linked Data.

The goal of PRISSMA is delivering an adaptive rendering engine for Linked Data resources. PRISSMA tweaks RDF visualization to the context in which the resource consumption is performed. Work in 2012 has been mainly focused on determining the algorithm that selects the best RDF visualization according to the real, sensed context. The uncertain and incomplete nature of context data, led to investigate strategies that model the task as an inexact RDF instance matching operation.

The second main research line carried out in 2012 led to the creation of Shi3ld, an access control framework for Linked Data SPARQL endpoints. Shi3ld authorization procedure and Access Policies, defined using Semantic Web languages only, have been enriched with the notion of mobile context, thus enabling context-based access control (e.g. geo-temporal authorization policies).

A collateral research line has been carried out as the follow-up of the 9th edition of the Summer School on Ontology Engineering and the Semantic Web (http://sssw.org/2012/ ). We contributed to an exploration work on the problem of spamming in Linked Data, providing a classification of potential spamming techniques and populating and publishing a dataset containing spammed triples(http://www-sop.inria.fr/members/Luca.Costabello/spam/ ). The dataset is useful to train anti-spamming mechanisms.