Discovering patterns of online popularity from time series

Abstract

How is popularity gained online? Is being successful strictly related to rapidly becoming viral in an online platform or is it possible to acquire popularity in a steady and disciplined fashion? What are other temporal characteristics that can unveil popularity of online content? To answer these questions, we leverage a multi-faceted temporal analysis of the evolution of popular online contents. Here, we present dipm-SC: a multi-dimensional shape-based time-series clustering algorithm with a heuristic to find the optimal number of clusters. First, we validate the accuracy of our algorithm on synthetic datasets generated from benchmark time series models. Second, we show that dipm-SC can uncover meaningful clusters of popularity behaviors in a real-world Twitter dataset. By clustering the multidimensional time-series of popularity of contents coupled with other domain-specific dimensions, we uncover two main patterns of popularity: bursty and steady temporal behaviors. Moreover, we find that the way popularity is gained over time has no significant impact on the final cumulative popularity.

Political polarization drives online conversations about COVID ‐19 in the United States

“…TA B L E 6 The temporal hashtag clusters detected by the dipm-SC method (Ozer et al, 2020), and the seman cally similar sub-clusters detected by the Louvain method (Blondel et al, 2008). TA B L E 7 The retweet likelihood ra o R(U i ← U j ) of hashtag sub-clusters i (rows) and j (columns) (see Eq.…”

“…We first set out to detect hashtag clusters from the me series of cluster usage by adop ng the recently-proposed mul dimensional me-series clustering framework named dipm-SC (Ozer et al, 2020), which is specifically designed to cluster social media me series. The shi window is set to three days for two reasons.…”

“…Second, a manual scan of the daily hashtags with the greatest IR values in the US suggests that some popular topics last at most a few days. To find the most op mal k, the number of clusters, following the dipm-SC framework (Ozer et al, 2020), we explore the temporal shapes of clustering results. We find that k = 4 produces the most non-overlapping temporal trajectory peaks.…”

“…F I G U R E 3 Popularity gain trajectories of the temporal hashtag clusters detected using the dipm-SC method (Ozer et al, 2020).…”

Twitter como recurso metodológico en Educación Superior: una experiencia educativa con estudiantes de Trabajo Social

“…Al respecto, Marín-Díaz y Cabero-Almenara (2019) reflexionan sobre la necesidad de generar más investigación científica que aporte modelos de referencia para la utilización de estas tecnologías en ámbitos universitarios. Otro aspecto fundamental a tener en cuenta es que, de inicio, existe una visión negativa generalizada por parte del profesorado y del alumnado para incorporar las redes sociales online como elemento metodológico en enseñanzas de Educación Superior (Andreev et al, 2020;Van Dijck, 2020), unido a una preocupación del usuario sobre cuestiones relacionadas con la seguridad, la privacidad o la falta de control sobre los mensajes emitidos en la red (Sai et al, 2020;Miller, 2020;Ozer et al, 2020).…”