BiasScanner: Automatic Detection and Classification of News Bias to Strengthen Democracy

TL;DR

BiasScanner tackles the challenge of helping news readers assess bias in online articles by delivering sentence-level bias detection and explanations through a browser plug-in backed by a server-side transformer classifier. It employs a GPT-3.5-turbo-16k model fine-tuned on the BABE dataset to identify 27 bias types, highlighting biased sentences and providing a structured bias report, all while emphasizing privacy by avoiding PII leakage. The system is designed for easy deployment and model interchangeability, with a web demo and Firefox extension available, and demonstrates competitive performance on established benchmarks (notably a high F1 around 0.76). Limitations include language scope, potential subjectivity in bias labeling, and reliance on proprietary back-end models; future work envisions open-source models, multilingual support, and broader content analysis such as hate speech and fake-news cues. Overall, BiasScanner presents a practical, privacy-preserving approach to aid democratic engagement by enabling real-time, transparent bias scrutiny in web news consumption.

Abstract

The increasing consumption of news online in the 21st century coincided with increased publication of disinformation, biased reporting, hate speech and other unwanted Web content. We describe BiasScanner, an application that aims to strengthen democracy by supporting news consumers with scrutinizing news articles they are reading online. BiasScanner contains a server-side pre-trained large language model to identify biased sentences of news articles and a front-end Web browser plug-in. At the time of writing, BiasScanner can identify and classify more than two dozen types of media bias at the sentence level, making it the most fine-grained model and only deployed application (automatic system in use) of its kind. It was implemented in a light-weight and privacy-respecting manner, and in addition to highlighting likely biased sentence it also provides explanations for each classification decision as well as a summary analysis for each news article. While prior research has addressed news bias detection, we are not aware of any work that resulted in a deployed browser plug-in (c.f. also biasscanner.org for a Web demo).

Figures (3)

• Figure 1: BiasScanner System Architecture
• Figure 2: In-line highlighting of biased sentences
• Figure 3: Bias Summary Report