The final objective of SENDUP is to propose a suite of software module to support the anonymisation of RDF/S databases. The targeted applications are:
Projecting RDF databases to reduce noise : Online anonymisation
Differential privacy can be achieved by adding random noise to the result of a query. The amplitude of said noise is typically calibrated depending on how much the output of a query vary between neighboring databases. Unfortunately, this variation can be high -or even unbounded- over the space of RDF graphs. This means that the randomness introduced by a differential private mechanism would be high, producing results that are not very usable. To compensate for this, we proposed a projection approach for typed and directed graphs onto bounded subspaces to reduce the magnitude of the noise.
GrAnon: Offline anonymisation module
GrAnon is an open-source engine supporting the specification and execution of sanitization procedures on attributed directed multi-graphs. GrAnon is based on a small langage of simple yet expressive operators formalized as graph rewriting rules.
It should be used to sanitized a database before its publication. Alternatively, the database can be kept unpublished and safely queried.
SetUp: Database evolution module
This software, called SetUp, manages evolutions of RDF/S databases. It takes as input a database and an update U (typically provided by the offline anonymisation module). SetUp verify the applicability of U and, if it cannot be applied without introducing an inconsistency, SetUp generates a set of updates called side-effects required to ensure U’s applicability. SetUp handles the application of the side-effects and, ultimately, U.