Kummer, MichaelAntunes, Frederico Ferreira Da Silva Ruivo2025-03-282025-03-282025-01-23http://hdl.handle.net/10362/181592In a digital landscape where capturing reader attention is crucial to news platforms, the Upworthy Research Archive enables the study of how article headlines influence engagement. This research assesses how the components of a headline impacts user clicks through machine learning methods to extract features related to emotional and phrase-meaning topics to predict the level of engagement on each piece of news. Results show that timing dominates engagement, headline structure plays a key role, and emotional traits have limited influence. Additional analysis explores how headline characteristics impacts Upworthy’s engagement.engDigital contentEngagementUpworthyMINDPredictionNews articlesHeadlineExcerptCategoriesClassification algorithmsMachine learningNatural language processingUser interactionContent strategyInteraction trendsPatternsInsightsText embeddingsForecastingTime seriesClickbait detectionHeadline classificationText analysisClickbait trendsPlotsStatistical analysist-Testt-TestHypothesis testingModel evaluationmaster thesis203927524