Dieses Dokuwiki verwendet ein von Anymorphic Webdesign erstelltes Thema.

# PhD position

## Visual Analytics for Multiplex Networks

Application deadline: (a soon as possible)

Host research institute : CNRS LaBRI UMR 5800, Models & Algorithms for Bio-informatics and Visualization team

### How to apply

Send your resume, intention letter and graduate studies record. Please clearly indicate why and how your profile is a good fit for the project. A description and pointers at past projects and realizations related to the PhD topic is a plus.

Academic advisor : Guy Melançon, Guy.Melancon@labri.fr
Université de Bordeaux / CNRS UMR 5800 LaBRI
Adresse : CNRS LaBRI UMR 5800
Campus Université Bordeaux I
351 Cours de la Libération
33405 Talence Cedex

Industrial advisor : Marie-Luce Viaud, mlviaud@ina.fr
Adresse : INA, 4 avenue de l’Europe
94360 Bry sur Marne

### Project abstract

Nowadays, information follows complex paths: information propagation involving on-line editors, 24-hour news providers and social medias follows entangled paths acting on information content and perception. Content digitalization and social networks have a clear impact on media actors which strive to analyze and understand these interactions between information mediators and their influence on information.

Multiplex networks are particularly good at modeling this context where citations and official information sources get mixed. These networks are multivariate and multilayers Combining numerous and distinct topologies into different layers where entities (nodes and edges) are equipped with several attributes (numerical, ordinal, semantic, etc.). The PhD project is an opportunity to extend standard approaches to multiplex networks in relation with the target application domain. It is also an opportunity to build visualizations combining different graphical representations of networks (coordinated multiple views as well as hybrid visualizations). The project should also focus on interactions that seem relevant to manipulate and navigate multiplex networks. We expect these visualizations to exploit statistics specific to multiplex networks; computing hierarchies for multiplex networks may also prove to be a fruitful avenue.

#### PhD profile

• Training : Engineering School or Master's degree in a related field
• Computer Human Interaction, Visualization
• Data mining
• Significant programming experience (Java, C/C++, python, javascript ; Agile Methods)
• 3 year contract, 34000 euros/yearly gross salary, yearly bonuses and salary increase.

#### Context and objectives

Nowadays, information follows complex paths that go far beyond the classical emitter-receptor model. While press editors and multimedia actors are still acting as information producers, they do not control the circuits along which information propagates anymore. Information propagation involves on-line editors, 24 hours news channel and social networks crossing multiple paths and acting on information content and perception. For instance, Iranian president Hassan Rouhani's declaration to CNN and reconciling messages published on his tweeter account helped the international community to build friendly diplomatic relations. Multiple heterogeneous sources converge and are propagated, and provoke diplomatic reactions turning news coverage into a complex set of more or less agreeing viewpoints. Medias are going through a deep change due to information digitalization and social networks, and must therefore analyze and understand these interactions between information mediators and their influence on information.

Multiplex networks are particularly good at modeling this context where the “circulatory circulation of information, as theorized by sociologist Pierre Bourdieu – pointing at how television channels played a copycat game, is amplified and turned into endless citations, comments and shared pieces of information.

The PhD project should contribute to:

• Favor the discovery, visualization and exploration of complex relationships supporting questions such as “What medias act as prescriber for specific events?” “Are social networks amplifying traditionnal medias or are they carrying genuine infomation?” “What sources do have authority?” “WHat impact does a series if events have?” (for instance, how is an international crisis as in Ukraine covered my the French medias and web?”
• Allow journalists and social sciences researchers to rescale things. Most studies only consider limited corpus both in size and time. The project adopts a Big Data perspective by combining media mining of the French web with social networks (Twitter, Facebook, identi.ca, …).

On the one hand, the project requires to show how multiplex networks help model the media environment. On the other hand, usable tools must be designed and realized to validate these models (and explore networks and simulate propagation).

#### Expected work

The complexity of the data we need to process comes not only from its massive volume, but mainly from its variety, heterogeneity and temporality. The data can be modeled using multivariate and multiplex networks: networks combining diverse topologies into layers where entities are equipped with attributes (numerical, ordinal, semantic, etc.) (Kurant & Thiran 2006) (Kivelä et al. 2013) (Schreiber et al. – to appear).

The PhD project is an opportunity to extend standard approaches to multiplex networks in relation with the target application domain. It is also an opportunity to build visualizations combining different graphical representations of networks (coordinated multiple views as well as hybrid visualizations). The project should also focus on interactions that seem relevant to manipulate and navigate multiplex networks. We expect these visualizations to exploit statistics specific to multiplex networks; computing hierarchies for multiplex networks may also prove to be a fruitful avenue.

The PhD project requires to develop new approaches to explore and navigate large networks of multimedia documents: multiscale modeling; classification/clustering and pre-processing of massive and complex data; computing representations matching INA's multi-faceted data; interaction well suited to content and multiscale models, and to tasks relevant to multimedia document archival.