Data Modeling for Data Governance
Corporations, government agencies and other organizations dedicate a huge amount of effort to defining document and data governance strategies to help contend with fast-growing and diverse pools of digital resources (e.g. data files, documents) in a variety of formats. At XRCE we have been working on information systems and models to ensure the coherent management of documents, and in general digital information, in continuously evolving settings. We are especially interested in treating data streams in a semantically coherent manner.
At the core of our work we propose a method that preserves a number of interdependent digital entities, including documents, in conformance with context related information. We assume such entities exist within a continually changing environment, which may result in them becoming inconsistent and unusable over time. A major aspect of the methodology is viewing the collection of digital entities as a network of resources where the links between them represent their dependencies (example Figure 1). A change that influences one of these objects can then be propagated to the rest of the objects via an analysis of the represented dependencies. We propose to model dependencies not only as simple links but as complex, semantically rich, constructs that encompass context-related information, as presented in  and . In this manner, dedicated change propagation algorithms take into account the context, encoded in dependency-specific properties. The context includes spatio-temporal and provenance related information (e.g. who created an entity and when) but also causal aspects including the notions of intention (the goal of a change) and invariance (the set of object features that should remain unchanged) (see Figure 1 ). This information has been formally described in an ontology that we call the Linked Resource Model (LRM) using Semantic Web standards and specifically the Web Ontology Language (OWL). This model has been developed within the context of the European FP7 project PERICLES.
To deal with the streaming related aspects of the above setting, we are developing an infrastructure and a novel query language that allows reacting to stimuli received from the environment in a streaming fashion. Reactions include performing local updates, but also sending and asking for information from other systems, waiting for responses, and requesting changes. Key technical elements of our work include the introduction of explicit operators to deal with concurrency, time, nested transactions, and streams in a Web context. Some of these ideas are described in .
Versioning is a form of change that is particularly challenging in dynamic, non-centralized architectures, as propagating changes in a consistent manner requires the strict coordination of tiers. an example of this is service-oriented architectures (SOA). We proposed an innovative Version Management method to consistently manage digital resources throughout their lifecycle. As part of the method we define how resources can be associated with logical specifications written in an extensible logical formalism understood and agreed by tiers. We also describe how a service can verify the well-formedness of proofs and properties when required, so that the version labels stay consistent. More information can be found at .
 Nikolaos Lagos, Jean-Yves Vion-Dury. Digital Preservation Based on Contextualized Dependencies. ACM International Symposium on Document Engineering, pp. 35-44, 2016.
 Jean-Yves Vion-Dury, Nikolaos Lagos, Efstratios Kontopoulos, Marina Riga, Panagiotis Mitzias, Georgios Meditskos, Simon Waddington, Pip Laurenson, Ioannis Kompatsiaris. Designing for Inconsistency - The Dependency-Based PERICLES Approach. ADBIS (Short Papers and Workshops), pp. 458-467, 2015.
 Jean-Yves Vion-Dury, Nikolaos Lagos. The Resource Action Language (ReAL): Towards Designing Reactive RDF Stores. SEMPER@ESWC 2016: 45-54. – Best paper award
 Jean-Yves Vion-Dury, Nikolaos Lagos. Semantic Version Management based on Formal Certification. ICSOFT-PT 2015: 19-30. – Best paper award