Section: New Results

Graph-based Knowledge Representation

Participants : Olivier Corby, Catherine Faron-Zucker, Fabien Gandon, Isabelle Mirbel, Adrien Basse, Oumy Seye.

We have designed a method to build pretty-printers for RDF Abstract Syntax Trees and Graphs, built on top of SPARQL Query Language. Pretty-print rules are written as SPARQL select-where queries. The where clause matches the target subtree to be printed and the select clause returns the pretty-printed statement using an external kg:pprint function. This function recursively calls the pretty printer, looking for appropriate pretty-print queries for the target subtrees.

We have designed a syntactic extension to SPARQL in order to ease writing pretty-printing rules. Below, an example of template for a SPIN like AST is shown:

template {

  "select " ?s

  "\n"

  "where "  ?w

}

where {

  ?in ast:select ?s ;

      ast:where  ?w

}

We were able to write a pretty-printer for a SPIN like complete SPARQL 1.0 AST with 19 templates.

We have designed an extension to our KGRAM SPARQL interpreter that enables to consider an RDF Graph directly as a Query Graph. This enables to compute RDF subgraph matching.

We have completed SPARQL 1.1 implementation with the final version of Property Path (PP) and federated queries (service & bindings). In order to be able to query a SPARQL 1.0 endpoint with PP, we have written a compiler that translates PP into basic graph patterns.

The work on KGRAM is published in [33] .

Alban Gaignard from the I3S Modalis team has designed a distributed version of KGRAM to query remote triple stores, in the context of Semantic Federation of Distributed Neurodata. This work is published in [63] , [64] .

We have implemented a prototype of C-SET Commutative Replicated Data Type for RDF in Corese with Pascal Molli and Luis Ibáñez from U. Nantes [69] .

Extracting Graph Patterns to Characterize RDF Data Sources

This work takes place in the PhD Thesis of Adrien Basse.

Many Semantic Web applications address the issue of integrating data from distributed RDF triple stores. There are several solutions for distributed query processing such as SPARQL 1.1 Federation, which defines extensions to the SPARQL Query Language to support distributed query execution. Such extensions make it possible to formulate a query that delegates parts of the query to a series of services, but one issue remains: how to automate the selection of RDF triple stores containing relevant data to answer a query. This is especially true in the context of the Linking Open Data project where numerous and very heterogeneous datasets are interlinked, allowing for interesting queries across several sources. To decompose and send queries targeting only relevant stores, we need a means to describe each RDF triple store, i.e. an index structure which provides a complete and compact description of the content of the RDF triple store.

To know the content of a RDF triple store, we proposed to use graph patterns as basic structures for index items. In this thesis we present an approach to extract these graph patterns from RDF triple store. For this purpose, we extend Depth-First Search coding (DFS) [104] to RDF labeled and oriented multigraphs and we provide a join operator between two DFS codes so as to sequentially build the different levels of the index structure.

Insertion or deletion of annotations in the triple store may cause changes to the index structure. To handle updates in triple store, we proposed a procedure to identify exactly the changes in the first level of the index structure and propagate them to the following levels. The DFSR (Depth First Search for RDF) coding makes it possible for us to efficiently manipulate graph patterns, but is difficult to read (succession of integer numbers). To facilitate the reading of our index structure, we propose a visualization user-interface and algorithms to turn a DFS code into a more legible format like RDF. Our algorithm relies on Corese/KGRAM [95] . We have tested our algorithm on many datasets. During the building of index structures we keep a set of data in order to help us to better understand the progress of our algorithm and improve it.

Rules for the Web of Data

This work takes place in the PhD Thesis of Oumy Seye.

We have characterized the subset of SPARQL that can be expressed in RIF and, conversely, we have searched for the maximal RIF dialect that can be expressed as SPARQL construct-where queries. This work is published in [71] , [72] were we present the implementation of a RIF dialect with a SPARQL Rule Engine in Corese/KGRAM.

We have designed online services for RIF-BLD parsers for presentation syntax and XML syntax (http://wimmics-ws.inria.fr/rifparser ). We have also done an online service for RIF-BLD translation into SPARQL and RDF (http://wimmics-ws.inria.fr/riftosparql ).