How can we measure biases in the media with the support of artificial intelligence? In this interactive webinar, a group of researchers will present a project that is addressing this question.
The project employs several machine learning techniques such as speech recognition and natural language processing to assess media bias in Spanish TV outlets. Moreover, the project seeks to understand how media bias can affect viewers’ perceptions about crucial issues such as voting intentions or main social concerns.
The webinar is organised by the Centre for Media Pluralism and Media Freedom team in cooperation with the Spanish research team of its flagship Media Pluralism Monitor project, led by Jaume Suau Martínez (Blanquerna, Universitat Ramon Llull).
Description of the project
The project consists of two different parts. In the first one, the authors seek to assess media bias for Spanish TV outlets. For that purpose, they develop a measure that abstains from ideological considerations. The second part consists of understanding how media bias can affect viewers’ perceptions about different issues (i.e. voting intentions or main social concerns).
Part 1: measuring media bias
The authors collect daily news data from the principal Spanish TV Channels using several machine learning techniques, such as speaker recognition, to separate the different stories. Afterwards, they develop a matching algorithm that uses NLP to group the stories across channels. The matching is flexible enough to group among common topics and leave channel-specific segments unmatched.
The final step is to compute a measure of bias based on three characteristics:
1. Salience: the relative position of each story within the daily outline
2. Coverage: the time devoted to each segment
3. Valence: the tone, assessed via sentiment analysis
Part 2: the effect of media bias on ideological perceptions
Using econometric techniques, the authors further study the different forces behind viewers’ ideological preferences and the information received from the media, counting on data from the Centro de Investigaciones Sociologicas (CIS) and audience estimates of the media outlets.
Manuel Lleonart-Anguix, PhD candidate in Economic Analysis at Universitat Autònoma de Barcelona and Barcelona School of Economics (IDEA Graduate Program). His main research interests are game theory and political economy. He uses network analysis in his research to understand the information effects in heterogeneous societies.
Sergi Quintana, PhD candidate in Economics at Universitat Autonoma de Barcelona, Barcelona School of Economics, and CSIC. His main research interests are in applied microeconomics and education economics. He is focused on the development of structural models and causal analysis within policy interventions.
Luis I. Menendez, PhD candidate in Economics at Universitat Autònoma de Barcelona and Barcelona School of Economics (IDEA Graduate Program). His main research interests are social networks, information economics, and political economy. He uses machine learning techniques to understand online coordination and the evolution of social protests.