Section: New Results

Querying Heterogeneous Linked Data

Recursive queries

P. Bourhis published a paper at IJCAI [17] in cooperation with the University of Dresden in Germany. There he developed a highly expressive Web query language of the Datalog family, for which static analysis problems such as query containment remain decidable.

In cooperation with Links' associated team in Oxford, P. Bourhis obtained an article at ACM TODS [5] , where he studies the access of hidden data by recursive queries.

V. Hugot, A. Boiret, and J. Niehren study monadic second-order logic for unordered trees with data constraints on siblings. This language can be used to define recursive queries and schemas on unordered data trees [13] . They study restrictions of the logics, for which the usual static analysis problems become decidable, and study the complexity of the decidable cases. This work was done in cooperation with Paris 7.

Schemas

I. Boneva and S. Staworko contribute at ICDT the RDF schema language SheX [22] , which they developed in cooperation with members of the W3C. The usual open world approach of RDF is schemaless in that the alphabets of RDF data are left open, so that data from different sources and with different alphabets can be unified. This raises serious problems for query writing and thus for linked data integration, since a query may become invalid when the alphabet changes. A SheX schema allows to express constraints on the alphabets, node labels and edge labels of RDF graphs, so that database queries become safe with respect to future changes without closing the alphabet. In a previous work the studied the case of XML data trees instead of RDF graphs [6] .

A. Lemay and J. Niehren propose sublinear algorithms in the style of probabilistic property testing for validating XML data trees with respect to DTD [20] .

P. Bourhis studies streaming bounded repair with respect to schema violations [8] . This work is done in a cooperation with the University of Bordeaux and the University of Santiago in Chile.

Provenance

P. Bourhis obtained an ICALP paper [11] in cooperation with Télécom ParisTech. They show how to propagate provenance information for monadic second-order logics on trees or tree like structures with polynomial data complexity. In their provenance framework, they can show how to generalize various aggregation tasks for monadic second-order logics, that were known to be solvable with polynomal data complexity before.

In a cooperation with Tel Aviv, P. Bourhis obtained a ACM CIKM paper [18] , where they show how to approximately summarize data provenance.

Data integration

In a cooperation with the University of Toronto, R. Ciucanu obtained a paper at PVLDB [4] on how to gain control over data integration evaluations. I. Boneva, A. Bonifati and R. Ciucaniu presented a paper on graph data exchange with target constraints [14] in the GraphQ workshop, and proved that query answering is intractable in this context.