A Multi-Dimensional Framework for News Quality Assessment Integrating Polysemy, Sentiment, Bias Detection

In today’s media landscape, the adoption of Natural Language Processing (NLP) methods is crucial for maintaining clarity and enhancing content quality. The paper is about a full-fledged analytics dashboard, called NewsChannel Pro, that is targeted at professional newsrooms and combines high-level polysemy recognition, sentiment detection, readability and bias recognition, as well as legal risk estimation. The system uses a multi-layered methodology based on WordNet built word sense disambiguation using Lesk algorithm, ensemble sentiment analysis using VADER and TextBlob as well as heuristic-based bias detection with optional NRCLex and Empath lexicons. One of the new contributions is the combination of audience impact modeling with confusion probability estimation of ambiguous words which allows one to clarify the content being presented. The system also has the automated news rewriting features to enhance readability and neutrality. The application of the proposed framework on experimental assessment of real-life news articles shows that it is effective in detecting ambiguous content and giving actionable advice to news editors. The dashboard has a holistic quality scoring system that is associated with human evaluation on the clarity and professionalism of the news.