Annonce en lien avec l’Action/le Réseau : aucun

Laboratoire/Entreprise : LIG
Durée : 3 ans
Contact : patrick.loiseau@univ-grenoble-alpes.fr
Date limite de publication : 2019-09-30

Contexte :
The 2016 United States presidential election was marked by an information war that took place into different social media platforms [9, 20]. Particularly, the election was marked by the abuse of targeted advertising on Facebook. For example, a group of Russian citizens and companies were indicted by U.S. authorities for trying to influence on the 2016 US election trough the Facebook Ads Platform [1, 6]. Since then social media platforms such as Facebook and Google released transparency platforms where they give access to interested parties to ads that are identified as ‘political’ by their platforms.

While this is an important step, in the dataset we collected using AdAnalyst (adanalyst.mpi-sws.org) we observed that
there are ads related to politics and important issues that are not labeled as such by Facebook. Hence, we believe it is important for independent parties to audit ads on Facebook as well.

Sujet :
The goal of the PhD thesis is to detect and study problematic ads in social media and assess the impact they have on users. The candidate is expected to contribute in two ways:

1. The first goal of the thesis is to propose algorithms to detect issue and political ads using machine learning tools such
as convolution neural networks and natural language processing tools. The student can thereafter focus on other kinds of problematic ads that promote for example bogus cures for diseases, anti-vaxxer blogs, or scammy financial services.

2. The second goal of the thesis is to design and perform experiments that can evaluate whether and to which extent people are influenced by these problematic ads that appear in their Facebook timeline. This is important as users have no control over what ads appear in their timeline, and users might be influenced by ads even if they do not click on them.

The student will be able to work with more than 200k real-world ads received by more than 1000 users we collected using our browser extension AdAnalyst (www.adanalyst.mpi-sws.org). Throughout the project the student will be able to familiarize himself with the online targeted advertising ecosystems, and apply machine learning techniques on real world data.

Profil du candidat :
Candidates should hold (or be about to get) a MSc degree in computer science and have:
• Strong coding skills.
• Experience in working with data.
• Strong motivation.
• Interest in the societal impact of data-driven systems.
• Interest in cognitive sciences and experimental research.

Formation et compétences requises :
Candidates should hold (or be about to get) a MSc degree in computer science and have:
• Strong coding skills.
• Experience in working with data.
• Strong motivation.
• Interest in the societal impact of data-driven systems.
• Interest in cognitive sciences and experimental research.

Adresse d’emploi :
Laboratoire d’Informatique de Grenoble
Batiment IMAG, 700 avenue Centrale, Domaine Universitaire
38400, Saint Martin d’Hères

Document attaché : 2019_PhD_ads_algorithms.pdf