Title [GV] ĐỀ 20 - A. ĐỀ THI.docx ✅

4 B. which themselves training on granular data, push backlist titles into unexpected afterlives and foreground “long-tail” revenue that previously went unnoticed C. themselves trained on granular data, push backlist titles into unexpected afterlives and foreground “long-tail” revenue that previously went unnoticed D. having been trainings on granular data, push backlist titles into unexpected afterlives and foreground “long-tail” revenue that previously went unnoticed Question 21. A. a wider pipeline yields a narrower canon unless publishers diversify inputs and, crucially, rethinks how value are measured B. a wider pipeline yields a narrower canon unless publishers diversify inputs and, crucially, rethink how value is measured C. a wider pipeline yields a narrower canon unless publishers diversify inputs and, crucially, to have rethought how value was measured D. a wider pipeline yields a narrower canon unless publishers diversify inputs and, crucially, rethink of how is value measured Question 22. A. having ever, without pushback do these innovations enter legacy workflows B. does ever, without pushback do these innovations enter legacy workflows C. have ever, without pushback do these innovations enter legacy workflows D. ever, if at all, without pushback do these innovations enter legacy workflows Read the following passage and mark the letter A, B, C, or D to indicate the correct answer to each of the questions from 23 to 30. The phrase “filter bubble” has become a rhetorical catchall for the anxieties attending personalized media. Yet careful audits of online behavior suggest a subtler terrain: many users traverse heterogeneous sources even while algorithms calibrate feeds for relevance. The specter of informational isolation, though not imaginary, is often overstated when compared with offline routines such as selective friendship networks or partisan news habits. Scholars further caution against conflating algorithmic curation with ideological indoctrination. While some clusters of users exhibit selfreinforcing exposure, largescale studies repeatedly find limited evidence that personalization alone seals audiences off from dissenting viewpoints. Instead, patterns of attention seem to hinge on motivation, media literacy, and the broader architecture of platforms where recommendation logics intersect with human choice. During election cycles and publichealth crises, these dynamics become more combustible. Rumor cascades travel swiftly across loosely connected communities, exploiting the very bridges that many fear have been severed. Countermeasures – labeling disputed content, downranking coordinated manipulation, and boosting authoritative outlets – can mitigate harms, but they also raise normative questions about legitimacy and transparency. If anything, the literature urges a recentering of institutional responsibility: platforms shape exposure not merely by predicting preference but by designing the frictions that enable deliberation – reporting tools, crosscutting prompts, and tempo controls for virality. Whether this curatorial power diffuses polarization or inadvertently entrenches it remains an empirical question, one sensitive to context, region, and regulatory climate. (Adapted from Reuters Institute (Oxford), “Echo chambers, filter bubbles, and polarisation – a literature review”, 2022) Question 23. The word audits in paragraph 1 is closest in meaning to _______. A. denunciations B. censures C. examinations D. encomiums Question 24. Which of the following is NOT mentioned in paragraph 1 as part of the comparison with online behavior? A. selective friendship networks B. partisan news habits C. algorithmic indoctrination programs D. heterogeneous source traversal Question 25. Which of the following best paraphrases the underlined sentence in paragraph 2? A. Although clusters of likeminded users do appear, comprehensive research generally indicates that personalization by itself rarely prevents audiences from encountering disagreement, even if pockets of reinforcement exist.