Mapping the Media Landscape: Predicting Factual Reporting and Political Bias Through Web Interactions
Dairazalia Sánchez-Cortés, Sergio Burdisso, Esaú Villatoro-Tello, Petr Motlicek
2024-10-28
Summary
This paper discusses a new method for evaluating the reliability of news sources by analyzing their web interactions to predict factual reporting and political bias.
What's the problem?
Assessing bias in news sources is important for ensuring that people receive accurate information. However, identifying biases like political leanings or misinformation can be difficult. Traditional methods often rely on expert analysis, which can be costly and time-consuming. There is a need for a more efficient way to evaluate news sources and their reliability.
What's the solution?
The authors propose an extension to a method that estimates the reliability of news outlets by examining how they interact online, specifically through hyperlinks. They test four different reinforcement learning strategies on a large graph of news media links to see how well they can classify sources based on factual reporting and political bias. Their experiments show significant improvements in identifying biases and factual accuracy, and they also create a large annotated dataset of news sources categorized by these labels.
Why it matters?
This research is significant because it provides a new approach to understanding media bias and reliability, which is crucial for consumers of news. By improving how we assess the trustworthiness of information, this method can help people make more informed decisions about what to read and believe, ultimately contributing to a better-informed public.
Abstract
Bias assessment of news sources is paramount for professionals, organizations, and researchers who rely on truthful evidence for information gathering and reporting. While certain bias indicators are discernible from content analysis, descriptors like political bias and fake news pose greater challenges. In this paper, we propose an extension to a recently presented news media reliability estimation method that focuses on modeling outlets and their longitudinal web interactions. Concretely, we assess the classification performance of four reinforcement learning strategies on a large news media hyperlink graph. Our experiments, targeting two challenging bias descriptors, factual reporting and political bias, showed a significant performance improvement at the source media level. Additionally, we validate our methods on the CLEF 2023 CheckThat! Lab challenge, outperforming the reported results in both, F1-score and the official MAE metric. Furthermore, we contribute by releasing the largest annotated dataset of news source media, categorized with factual reporting and political bias labels. Our findings suggest that profiling news media sources based on their hyperlink interactions over time is feasible, offering a bird's-eye view of evolving media landscapes.