Polarity Boundary Innovation Resolves Zero Inflation In Journalistic Sentiment Analysis

AbstractGeneral Background Evaluating public satisfaction with executive leadership serves as an essential mechanism for ensuring democratic accountability. Specific Background Conventional public opinion surveys demand extensive resources, while modern web scraping encounters formidable anti-bot security systems and commercial data restrictions. Knowledge Gap Current analytical prototypes frequently apply arbitrary limits causing zero-inflation anomalies where all formal media texts are misclassified as neutral alongside cross-language lexicon disparities. Aims This study develops a multi-regional news syndication framework utilizing TextBlob to calibrate classification parameters and profile semantic density. Results Processing coverage regarding the Indonesian presidency demonstrated that shifting the initial threshold successfully resolved extreme classification dominance, yielding a realistic distribution of positive, neutral, and negative semantic representations. Novelty This research formulates a specialized computational framework precisely adapted for formal publications to overcome extreme categorization anomalies in political science. Implications The proposed architecture delivers an efficient exploratory monitoring instrument for national leadership evaluation and establishes a methodological foundation for future localized deep learning integration. Highlights Processing syndicated digital feeds provides a cost-efficient alternative to conventional opinion polling. Modifying arbitrary classification parameters significantly reduces computational errors across formal news datasets. Lexical evaluation reveals dominant public discourse centered on executive administration programs. Keywords Natural Language Processing; News Syndication; Threshold Calibration; Public Satisfaction; Political Communication