Martin Gueuning (UNamur)

Title: Information Diffusion on Twitter: exploiting the timings of Retweets

Abstract: Online social networks such as Twitter have emerged as influential platforms for real-time information diffusion. Due to various factors, a tweet may become viral and initiate a cascade of retweets. In this work, we first show how the timings of the retweets reveal information about how the tweet propagated on the underlying network. In particular, we show under which circumstances the transition of the tweet from one group to another on Twitter leaves fingerprints on its retweets timings [1]. Then, we leverage these findings to address the question of multi-targets selection for maximal information diffusion. We present SmartInf, an algorithm providing a list of influential users of the network only based on an historic of popular cascades, or refining the list when the ego networks of the users may be extracted. We illustrate the performance of SmartInf against the baselines on both real-world and simulated data [2] .

[1] Temporal patterns of (re)tweets reveal cascade migration. AK Bhowmick, M Gueuning, JC Delvenne, R Lambiotte, B Mitra. Proceedings of the 2017 IEEE/ACM International Conference on Advances in in Social Networks Analysis and Mining.
[2] Temporal patterns of (re)tweets help to detect influential nodes for information diffusion in social networks. AK Bhowmick, M Gueuning, JC Delvenne, R Lambiotte, B Mitra. Submitted.