Section: New Software and Platforms

GRAAL

In its current state, Graal allows storage of data via a generic interface in different storage paradigms and systems. Currently, the relational database management systems MySQL, PostgreSQL, Sqlite, and InMemory graph and LinkedList structures are implemented. The triple store Jena TDB and the graph database system Sparksee are coming soon. Graal also allows us to query this database taking into account an ontology represented by a set of existential rules. It provides forward chaining and query rewriting algorithms (building up on Mélanie König's PhD thesis) and a tool for the analysis of the properties of a set of rules which is an integration of Swan Rocher's tool Kiabora. The input and output of this software can be expressed in our Datalog-inspired format DLGP 2.0 ( 6.1 ), and can be translated from the semantic web language OWL2 or to RuleML. This software is designed in a modular way, hence it is possible to use only a subpart of Graal without embedding it all or to easily replace an implementation of a module by another.

Functional Description

Graal is intended to be a generic platform for ontology-based query answering with existential rules.

• Participants: Clément Sipieter, Swan Rocher, Jean-François Baget, Marie-Laure Mugnier, Michel Leclère

• Partner: LIRMM

• Contact: Marie-Laure Mugnier (scientific contact) and Clément Sipieter (technical contact)

• URL: http://graphik-team.github.io/graal/

Objectives: Graal is a generic platform for query answering under existential rules. It will integrate all algorithms designed in the team, and our ambition is to make it a reference platform in the research community, allowing for the integration of algorithms designed by other teams.

Users community: Graal is intended for use in research and education.

Impact: Due to the recent release of the first stable version, Graal has only been used for now in our projects. A related paper received the RuleML 2015 challenge award http://2015.ruleml.org/ .

State of the art: To the best of our knowledge, the only other tool for reasoning with existential rules is Nyaya (De Virgilio, R.; Orsi, G.; Tanca, L.; Torlone, R., "NYAYA: A System Supporting the Uniform Management of Large Sets of Semantic Data," in Data Engineering (ICDE), 2012 IEEE 28th International Conference on , vol., no., pp.1309-1312, 1-5 April 2012), a joint development from teams in Rome, Oxford, and Milan. It has been renamed IRIS+/- https://bitbucket.org/giorsi/nyaya .

Misc.: Graal is written in Java (around 30k lines of code, 30 man/months effort). It is mainly developed by Clément Sipieter (2 years Inria ADT funding) under the CeCILL licence (GPL compatible, see http://www.cecill.info/licences.fr.html ). The development started 1.5 years ago from a prototype realized during Bruno Paiva's PhD thesis, and integrates work carried out by other PhD students (Mélanie König and Swan Rocher). Graal has been first presented at RuleML 2015 [23] , [33] , where it received a best paper award.

New features: Main features integrated in 2015 are query rewriting algorithms, projection algorithms, and translations to and from other languages (OWL2, RuleML).

Note that we do not detail here other software developments internal to our current projects and not publicly available.