PhD position at Université de Lyon (LIRIS) with Métropole de Lyon

Title: Defining a sustainable workflow for semantically enriching
territorial Open Data

Context:

Like many other territorial collectivities, the Métropole de Lyon [1] is
publishing as Open Data several datasets that are produced either by
the Métropole
de Lyon itself or by its partners [2]. As is often the case, these datasets
are available in relatively standard formats (e.g. CSV, JSON, GeoJSON,
...), providing a good level of syntactic interoperability. Yet, semantics
is quite shallow. The meaning of data is often conveyed out of band,
through textual documentation in the metadata as well as private
communications between data consumers and the data.grandlyon.com help desk.
Moreover, sometimes consumers rely on their own interpretations, which
certainly may be erroneous. As such, much of the potential value of the
published Open Data is hard to exploit:


   -

   the semantics of each dataset has to be hard-coded into each client
   application,
   -

   making these applications prone to break each time the underlying data
   evolve,
   -

   preventing data consumers from easily building value-added applications
   mixing data from multiple datasets.


Linked Data [3] is a set of technologies and standards recommended by the
W3C, aiming to solve the problems above. Such technologies rely on a
graph-oriented data model (RDF), which  facilitates data integration and
fosters the use of shared vocabularies equipped with explicit semantics.
The extra effort required from data publishers to comply with Linked Data
principles benefits all the consumers, as it renders each dataset easier to
use and  to link with other datasets, the latter being even more important.

The Métropole de Lyon plans to have its Open Data progressively comply with
Linked Data principles. Such a long term goal provides the general context
of this PhD proposal.

Topic:

A number of languages and tools have already been proposed, allowing to
migrate legacy data to Linked Data,  e.g.: R2RML [4], RML [5] and
“Plate-forme Territoire” [6], the latter being explicitly dedicated to
territorial Open Data. As for vocabularies, the growing interest for Linked
Data has lead to the publication of many vocabularies and ontologies (cf.
the LOV directory [7]), usable in different domains including those
relevant to territorial Open Data.

The first task of the PhD student will consist in reviewing the existing
tools and vocabularies, in order to single out those ones being  the most
suitable for the Open Data published by the Métropole de Lyon. Potentially,
such a task will call for an extension of the existing solutions.

The second step will consist in making sure that the migration workflow is
viable and maintainable in the long run. Indeed, semantic integration is a
continuous, never-ending process:


   -

   during the definition of the mapping between legacy data and some target
   linked vocabularies, misinterpretations can easily occur. When discovered,
   they obviously require an update of the mapping.
   -

   Datasets are updated with different frequencies, ranging from yearly to
   real-time. Of course, semantics should follow any evolution in the data
   production/update workflow, which may require an update of the mapping.
   -

   The way data are used in practice may reveal some subtle semantic
   aspects, not initially taken into account, which may also necessitate some
   update of the mapping.


Therefore, the proposed migration process must be robust with respect to
changes in the source data. It must also enable the staff of the Métropole
de Lyon to identify the evolutions to be applied to the entire workflow and
the best way to implement them.

The Phd student:

will work in Lyon (France), part-time at the Métropole de Lyon and
part-time at the LIRIS laboratory (UMR 5205 CNRS). S⋅he should have a
master degree or equivalent in computer science, with a strong background
in Web technologies and Knowledge representation. Ideally,  s⋅he would also
have some experience in Semantic Web and Linked Data. Good communication
skills, especially in English (written and oral) are also required. Some
knowledge of the French language would be a plus, but is not a strict
requirement.

Contacts:

Alessandro Cerioni <acerioni@grandlyon.com>

Pierre-Antoine Champin <pierre-antoine.champin@liris.cnrs.fr>


[1] https://www.grandlyon.com/

[2] https://data.grandlyon.com/

[3] https://5stardata.info/en/

[4] https://www.w3.org/TR/r2rml/

[5] http://rml.io/

[6] http://territoire.emse.fr/

[7] https://lov.linkeddata.es/