Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Self-Censorship on Facebook

Self-Censorship on Facebook

2014-01-08 Lab Seminar (Gais Lab)

Chaowen Chang

January 08, 2014
Tweet

More Decks by Chaowen Chang

Other Decks in Research

Transcript

  1. Self-Censorship on Facebook Sauvik Das and Adam Kramer
 International Conference

    on Weblogs and Social Media 2013 Kojima@CCU CSIE Undergraduate
  2. Introduction • Self-censorship (⾃自我審查) • Self-censorship and Last-minute self-censorship •

    An action: filtering • It’s both helpful and hurtful. why? • Conditions
  3. Introduction cont. (1) • 部分在 Facebook 中的 390 萬使⽤用者⾏行為分析 •

    了解⾃自我審查在其中的規模 • 識別⾃自我審查在不同類型區域中的形態
 (群組、個⼈人動態或評論) • 根據⽤用⼾戶的⾏行為去建構⾃自我審查在⼈人⼝口統計學 (demographic) 和社交圖譜 (social graph) 中的模型。
  4. Introduction cont. (2) • 390萬⽤用⼾戶中,71%的使⽤用者在 17 天的內對⽂文章或評論 有過⾄至少⼀一次的⾃自我審查。 • ⽂文章

    ⽐比 評論 更常⾒見此⾏行為。 
 (Posts 33% vs. Comments13%) • 使⽤用者會強烈地受到”預期的觀眾”影響⽽而決定發不發⽂文。
  5. Introduction cont. (3) • 男性審查較⼥女性多。 
 (但通常是男性有較多同性朋友的時候) • ⼈人們有較多限制時會更常⾒見此審查⾏行為。 •

    舊有使⽤用者較新使⽤用者審查的少,
 但是舊有使⽤用者審查評論⽐比⽂文章還多。 • 在朋友們的政治傾向和年齡分部變化較多的⼈人會審查的更 少。
  6. • 本⽂文將這 390萬 Facebook 使⽤用者 17 天內的 ⾃自我審查
 (考慮過,但最終仍沒有發表) 


    歸類為以下五種原因: • 不想要煽動或是繼續論證 (筆戰?) • 不想得罪他⼈人 • 不想煩他⼈人 (造成他⼈人困擾) • 不想發表會使他⼈人覺得⾃自⼰己表現不⼀一致的內容 • 技術限制 (例如:⼿手機、平板上發表不易⽽而中斷取消) Related Work
  7. Methodology cont. (1) • 這裡將 “Self-censorship” ⾏行為定為
 「使⽤用者寫了些東⻄西,但最終並沒有發表」 • 本質即是:「使⽤用者寫了東⻄西就是有意圖要分享,但最終

    卻決定不分享出來,故認為其進⾏行了審查。」 • 可涵蓋多數⾃自我審查,但不能包含全部得⾃自我審查,
 僅能計算那些在最後⼀一刻選擇放棄發表的使⽤用者。
  8. Methodology cont. (2) • To mitigate noise in our data,

    content was tracked only if at least five characters were entered into the composer or comment box. • Content was then marked as “censored” if it was not shared within the subsequent ten minutes. • Using this threshold allowed us to record only the presence or absence of text entered, not the keystrokes or content.
  9. Methodology cont. (3) • These analyses were conducted in an

    anonymous manner, so researchers were not privy to any specific user’s activity. • Furthermore, all instrumentation was done on the client side. • In other words, the content of self-censored posts and comments was not sent back to Facebook’s servers: Only a binary value that content was entered at all.
  10. Methodology cont. (7) • H1: posts will be censored more

    than comments. • H2: men will self-censor more than women. • H3: users with more opposite-sex friends will self-censor more. • H4: younger users will self-censor less. • H5: users with older friends will censor more. • H6: users who more frequently used audience selection tools self-censor less. • H7: users with more diverse friends will self-censor more. Hypothesis
  11. Results cont. (3) • 幾乎全部符合 H6: users who more frequently

    used audience selection tools self-censor less. • 唯⼀一例外在群組動態中的⽂文章發佈⾃自我審查很頻繁。 • ⽽而在評論中,對於相⽚片內評論的審查 (15%) 和 群組內評 論 (14%) 較 時間軸和其⾃自⾝身動態更新還⾼高。 • 對相⽚片提意⾒見可能較容易引發問題使⼈人憂慮是合理的。
  12. Results cont. (5) • ⽂文章⽐比評論有較⾼高的機率被審查。 (33% vs. 13%)
 H1: posts

    will be censored more than comments.
 • 男性較⼥女性⾼高,但當男性其男性朋友⽐比例增加時會審查更多⽂文 章。 (對於評論則未觀察到受性別影響) • H2: men will self-censor more than women.
 在⽂文章部分成⽴立,但評論部分則否。 • H3: users with more opposite-sex friends will self-censor more.
 不論是⽂文章或評論皆不成⽴立。
  13. Results cont. (6) • 對於⽂文章,年⻑⾧長使⽤用者的⽂文章審查較使⽤用者為少,但在評 論則為多。 • H4: younger users

    will self-censor less.
 因此在⽂文章部分不成⽴立,但評論部分成⽴立。 • 審查與否朋友們的平均年齡無關。 • H5: users with older friends will censor more.
 皆不成⽴立。
  14. Results cont. (7) • 特別的是,使⽤用者參與越多群組會審查越多的⽂文章與評論。 • 同樣地,使⽤用選擇分享⼯工具(挑選觀眾)的使⽤用者會也會審查 更多的⽂文章與評論。 • 值得注意的例外是使⽤用私⼈人訊息的趨勢與評論的較低審查率

    有關。 • 儘管如此,針對特定群體的⽂文章和評論都⼤大幅增加了審查率
 H6: users who more frequently used audience selection tools self-censor less.
 特定群體的審查會增加,⽽而⾮非減少,故 H6 不成⽴立。
  15. Results cont. (8) • H7: users with more diverse friends

    will self-censor more.
 是綜合的結果,不⼀一定有較⾼高多元性就⼀一定審查較多。