Seminar by Noorani Bakerally

"Generation of Linked Data Platforms from existing data sources in highly decentralized information ecosystems" by Noorani Bakerally

at 1:30 PM

Room F021a
Building F
Laboratoire Hubert Curien
18 rue du Professeur Benoît Lauras
42000 Saint-Etienne

Seminar by Noorani Bakerally, ARMINES

Abstract

Information ecosystem with decentralized architectures have known some success in terms of data interoperability. However, today, information ecosystems in which organizations are operating are moving towards highly decentralized architectures for reasons such as globalization, multinationalism or complex supply chain collaborations. The problem in highly decentralized information ecosystem is that while data is heterogeneous, there is little to no coordination between its different units.
Consequently, resolving data heterogeneity and enhancing interoperability between the information systems become more challenging in such a context.
In this work, our hypothesis is that Semantic Web technologies and in particular the Linked Data Platform 1.0 (LDP) standard can be used to provide a homogeneous view and access to self-described data. Thus, it decreases the need for contacting data providers to understand and make use of the data. However, while several LDP implementations exist, they provide no support for automating the generation of LDPs. Consequently, deploying LDPs from existing data sources still involves a lot of manual development. To solve this problem, we propose an approach that uses a language to describe how existing data sources can be used to generate LDPs compatible with any implementation of the LDP standard and deployable on any of them. We formally describe the syntax and semantics of language. We provide an implementation of the approach that instantiates an automatized generation and deployment workflow. Finally, we evaluate our language and approach in general by performing several experiments to show how our approach automatizes the generation of LDPs, while enhancing design reusability, from existing data sources that are either heterogeneous or have hosting constraints.

This seminar will be held in english.