Influence and Diffusion-Aware Group Recommendation in Social Media

par Yangke Sun

Projet de thèse en Informatique

Sous la direction de Bogdan Cautis et de Silviu Maniu.

Thèses en préparation à Paris Saclay , dans le cadre de École doctorale Sciences et technologies de l'information et de la communication (Orsay, Essonne ; 2015-....) , en partenariat avec LRI - Laboratoire de Recherche en Informatique (laboratoire) , LaHDAK - Données et Connaissances Massives et Hétérogènes (equipe de recherche) et de Université Paris-Sud (établissement de préparation de la thèse) depuis le 01-10-2018 .

  • Titre traduit

    Influence and Diffusion-Aware Group Recommendation in Social Media


  • Résumé

    Recommendation is an ubiquitous task on the Web, where users are recommended items to adopt. In the common scenarios that are targeted by recommendation systems, people have always been influenced by their social circles to various degrees, by word-of-mouth. But this phenomenon has taken a whole new dimension with the advent of social networking on the Web. With a simple click, re-tweet, or “like”, one can now instantly exert influence and diffuse information to her peers. Therefore, a deeper understanding of the way items are diffused, endorsed, and used can greatly improve the effectiveness of recommendations. However, due to the highly complex and uncertain nature of data in real-world social media, this raises new conceptual and technical challenges. Better adapted data models and algorithms are needed to address them, for a new generation of recommendation mechanisms, which can be influence-aware and social network-aware. In this context, the PhD project aims at advancing the state-of-the-art in the area of group recommendation techniques, by taking into account key ingredients such as influence, diffusion patterns, and the community-based, modular structure of social networks. Group recommendation is important in many scenarios, such as those targeting movies, travel destinations, restaurants, online games, music playlists, etc. Our goal is to devise new models and algorithms, and to empirically validate them in diverse applications.