引導式測驗 · 中等 等級

過濾氣泡和推薦演算法測驗

測試過濾氣泡定義、經驗證據、研究結果、平台和使用者效果以及推廣研究結果的限制。

問題 1 的 8

本指南中的「過濾氣泡」通常指什麼?

預覽每個問題8 問題

當您參加測驗時,就會出現答案和解釋。

  1. 本指南中的「過濾氣泡」通常指什麼?
    • A technical method that removes all personalized recommendations
    • A synonym for any dislike of a video
    • A concern that personalization may narrow a person’s exposure to different information over time
    • A guarantee that platforms show identical content to every user
  2. 2015 年 Facebook 研究比較了什麼?
    • Diverse news shared in networks, content appearing in feeds, and what users chose to consume
    • Only the number of posts users created
    • Only changes in platform subscription prices
    • Political beliefs before and after a randomized election
  3. PNAS 在客製化研究平台中使用 YouTube 衍生建議進行的研究發現了什麼?
    • It proves that no platform can create filter bubbles
    • It establishes effects for every user over many years
    • It shows that user choices have no influence
    • It found limited short-term polarization effects under its tested recommendation conditions
  4. 為什麼「曝光」、「點擊」和「信念改變」不能互換?
    • Belief change can be inferred from the ranking code alone
    • They describe different stages and outcomes that require separate evidence
    • A click always proves a belief changed
    • Exposure is identical to engagement in every study
  5. 哪些設計細節會影響推薦效果的結論?
    • Only the color of the recommendation card
    • The file format used for study notes
    • The name of the ranking model alone
    • The content available to the system and the time horizon of the study
  6. How can a researcher separate ranking effects from user choice?
    • Assume all available items were displayed
    • Use watch time as the only measure of exposure
    • Measure what was eligible, what was shown, and what users selected
    • Count only the items a user liked
  7. What conclusion should be avoided from a short exposure study?
    • A note that researchers used real YouTube videos and perturbed YouTube-derived recommendations in a custom platform
    • A broad claim that the experiment proves long-term attitude changes on every platform
    • A statement about the measured short-term exposure outcome
    • A description of the platform and tested population
  8. Which overall conclusion matches the evidence described here?
    • Recommendation systems may narrow or widen exposure depending on design and context
    • Personalization always isolates users from opposing views
    • Personalization never affects what users see
    • A single universal effect applies across all platforms