Database sanitization allows to share and publish open (linked) data without jeopardizing privacy. During their sanitization, graph databases are transformed following graph transformations that are usually described informally or through ad-hoc processes.
GrAnon is the result of a first effort toward bridging the gap between rigorous graph rewriting approaches and graph sanitization by providing basic generic graph rewriting operators to serve as a basis for the construction of sanitization mechanisms.
Design principle
GrAnon relies on 8 basic operators, each formalized as a single graph rewriting rule following the Single Push-Out approach. Together, they can be combined into procedures expressive enough to provide local differential privacy guarantees and sensitive attribute anatomization.
Operators are invoked through a language describing sets of nodes or edges as the scope of our operators. This can be intuitively seen as clauses:
– WHERE allows restriction of the operator to a subset of nodes defined by a relation (e.g. nodes that share a type or specific attribute).
– EXCEPT allows exclusion by the operator of a subset of nodes or relations defined by a relation.
Code and usage
GrAnon can be used as a library through its API, through its textual user interface or through its graphical user interface. In addition to a parser and the atomic operator, the current version of GrAnon propose 2 built-in sanitization procedures that can be used to achieve local differential privacy and anatomization.
GrAnon is coded in Java. Operators and all related rewriting rules are implemented using the Attributed Graph Grammar System (AGG), providing a concrete tool implementing formal graph rewriting mechanisms to sanitize semantic graph databases.
GrAnon's user manual is available here and its sources are on github.
References
[1] Adrien Boiret, Cédric Eichler, Benjamin Nguyen. "Privacy Operators for Semantic Graph Databases as Graph Rewriting". DOING@ADBIS 2022 : 3rd International Workshop on Intelligent Data – From Data to Knowledge. 2022.
[2] Adrien Boiret, Cédric Eichler, Benjamin Nguyen, Sara Taki. "Graph Rewriting Primitives for Semantic Graph Databases Sanitization". Under review for the journal Computer Science and Information Systems.