Improved Models for Media Bias Detection and Subcategorization
This work addresses media bias detection for news analysis, but it is incremental as it builds on existing transformer methods with new data and techniques.
The paper tackled the problem of detecting and sub-classifying media bias in English news articles by comparing zero-shot versus fine-tuned transformer models and using synthetic data, resulting in improved performance on a novel taxonomy of 27 bias types.
We present improved models for the granular detection and sub-classification news media bias in English news articles. We compare the performance of zero-shot versus fine-tuned large pre-trained neural transformer language models, explore how the level of detail of the classes affects performance on a novel taxonomy of 27 news bias-types, and demonstrate how using synthetically generated example data can be used to improve quality