An Analysis of Event-Agnostic Features for Rumour Classification in Twitter

Abstract

Recently, much attention has been given to models for identifying rumors in social media. Features that are helpful for automatic inference of credibility, veracity, reliability of information have been described. The ultimate goal is to train classification models that are able to recognize future high-impact rumors as early as possible, before the event unfolds. The generalization power of the models is greatly hindered by the domain-dependent distributions of the features, an issue insufficiently discussed. Here we study a large dataset consisting of rumor and non-rumor tweets commenting on nine breakingnews stories taking place in different locations of the world. We found that the distribution of most features are specific to the event and that this bias naturally affects the performance of the model. The analysis of the domain-specific feature distributions is insightful and hints to the distinct characteristics of the underlying social network for different countries, social groups, cultures and others.

Birds of a feather check together: Leveraging homophily for sequential rumour detection

“…Despite increasing interest in analysing rumours in social media [12,4,13,14,15,16,11,17,18] and the building of tools to deal with rumours that had been previously identified [19,20], there has been very little work in automatic detection of newly-emerging rumours [21,22], i.e. rumours that were not observed in the training data.…”

Mentiras, prejuicios y rechazo. El uso de la desinformación para propagar discursos de odio contra personas migrantes y refugiadas en redes sociales: un estudio articulado del caso español

“…-El desarrollo de soluciones tecnológicas para la identificación de contenidos e imágenes digitales. En este área han sido numerosos los intentos para detectar de manera automática los contenidos falsos (por ejemplo, Dey et al, 2018;Kim et al, 2018;Pérez-Rosas et al, 2018;Tschiatschek et al, 2018) o los rumores (Zhao et al, 2015;Zubiaga et al, 2017;, si bien los resultados son todavía limitados, puesto que la detección casi siempre gira en torno a cuestiones estilísticas (Shu et al, 2017) y dado que las características de los estos contenidos cambian radicalmente entre eventos (Tolosi et al, 2016). Esta aproximación se desarrollará en mayor detalle en el Estudio 3.…”