Information Diffusion-Based Social Recommender Systems – running

The goal of recommender systems is to provide items (products, services) to users, which they will probably like in the future. Traditional recommender systems only use the information of the users' past behaviour and do not take into account the users' social relations. However, in real life, people often ask their friends, colleagues to recommend a movie to watch, a restaurant to go to etc. That is, people's decisions are influenced by their social connections. As opposed to traditional recommender systems, social recommender systems incorporate the information of the people's social network in the recommending process to enhance the accuracy of predictions.

In this project we build social recommender systems that use existing, well-known information diffusion models to provide personal recommendations for users. These information diffusion models are used to describe and model the propagation of information (e.g. users' opinions, decisions) in the users' social network, based on which we provide predictions for the users' future decisions. However, the drawback of these recommender systems is that the number of users and the set of data to be propagated in the social network are enormously huge. That is, running the information diffusion models becomes a very resource-intensive task.

Project owner:
Pósfai Gergely (Távközlési és Médiainformatikai Tanszék)
Távközlési és Médiainformatikai Tanszék (VIK-TMIT)