An Evaluation of Structural Characteristics of Networks to Identify Media Bias in News Portals
Resumo
Nowadays, news websites are the main sources of information for most people. But these outlets may have a bias in their publications, favoring some ideology. This can influence people's opinion regarding several topics. Methods to automatically detect bias are usually based on aspects such as text and embedded hyperlinks. In this work, we present a strategy to detect media bias by analyzing networks that model citations between news websites, evaluating if the bias is related to structural characteristics of these networks. We show that ideological bias can be captured by citation patterns and present a method that uses these patterns to automatically identify political bias in news websites, reaching accuracy and F1 scores above 72% and 0.70, respectively.
Palavras-chave:
media bias detection, news analysis, network analysis, online news
Publicado
30/11/2020
Como Citar
AIRES, Victória Patrícia; SILVA, Altigran Soares da; NAKAMURA, Fabíola Guerra; NAKAMURA, Eduardo Freire.
An Evaluation of Structural Characteristics of Networks to Identify Media Bias in News Portals. In: BRAZILIAN SYMPOSIUM ON MULTIMEDIA AND THE WEB (WEBMEDIA), 1. , 2020, Evento Online.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2020
.
p. 172-179.
