A time-sensitive model to predict topic popularity in news providers¹

Article type: Research Article

Authors: Prado-Romero, Mario Alfonso^{a; *} | Coto-Santiesteban, Alex^b | Celi, Alessandro^c | Stilo, Giovanni^c

Affiliations: [a] University of Havana, Havana, Cuba | [b] Gran Sasso Science Institute, Italy | [c] University of L’Aquila, Italy

Correspondence: [*] Corresponding author: Mario Alfonso Prado-Romero, University of Havana, Havana, Cuba. E-mail: mario.prado@matcom.uh.cu.

Note: [1] This is an extended and revised version of a preliminary conference report that was presented in CIARP 2019 [22].

Abstract: The volume of news increases everyday, triggering competition for users’ attention. Predicting which topics will become trendy has many applications in domains like marketing or politics, where it is crucial to anticipate how much interest a product or a person will attract. We propose a model for representing topic popularity behavior across time and to predict if a topic will become trendy in the future. Furthermore, we tested our proposal on a real data set from Yahoo News and analyzed the performance of various classifiers for the topic popularity prediction task. Experiments confirmed the validity of the proposed model.

Keywords: Collective attention, predictive model, topics’ popularity model, machine learning

DOI: 10.3233/IDA-200012

Journal: Intelligent Data Analysis, vol. 24, no. S1, pp. 123-140, 2020