114 年 國立臺灣大學圖書資訊系碩士班甲組《專業英文（J）》
第 1 題
請閱讀以下短文,並以英文簡答(40%):
Scientific reproducibility, a cornerstone of research validation, has become a growing concern in the academic
community. A comprehensive study surveying approximately 1,500 scientists revealed a widespread challenge:
more than 70% of researchers failed to reproduce experiments conducted by their peers, while over 50%
encountered difficulties replicating their own previous work. This apparent crisis in methodological consistency
has sparked debate within the scientific community. Interestingly, despite these reproducibility challenges, most
researchers maintained confidence in published literature, with 73% believing that at least half of the papers in
their field could be trusted. Only 31% of scientists considered failed replication as an indicator of invalid results.
The study identified several key factors contributing to reproducibility problems. The most significant issues were
publication bias and selective reporting, where scientists tend to report only successful results. Other factors
included insufficient testing of experiments, inadequate peer review, and technical difficulties. Communication
was also a major barrier - less than 20% of researchers had been contacted by others about failures to reproduce
their work.
Different scientific fields showed varying levels of commitment to addressing these challenges. Medical research
institutions led the way, with 41% of laboratories implementing specific measures to improve experimental
reproducibility. In contrast, only 24% of laboratories in physics and engineering fields adopted such measures.
Most scientific institutions suggested specific improvements, including better experiment design, enhanced data
analysis, improved research training, and standardized testing methods. While implementing these improvements
could increase research time by 30%, 80% of scientists believed such changes were necessary.
ATTN: Use specific examples and statistics from the passage to support your answers.
- What is "reproducibility" and how many scientists in the survey had problems with it? Support your
answer with numbers from the passage. (8%) - The passage reveals a contradiction in scientists' attitudes regarding the notion of reproducibility. Explain
this contradiction using statistics from the passage. (8%) - According to the passage, what are the barriers to improving scientific reproducibility? Identify at least
two barriers and explain their potential impact on scientific research. (8%) - According to the passage, what differences exist between medical research and physics/engineering fields
in addressing reproducibility? Include specific percentages in your answer. (8%) - What specific improvements does the passage suggest for enhancing reproducibility? (8%)
登入後即可作答並保存紀錄。
核心觀念
本題出自國立臺灣大學圖書資訊學研究所（乙組）專業英文考科，測驗考生對**學術傳播（Scholarly Communication）與開放科學（Open Science）**核心議題「學術研究可重複性／再現性危機（Replication / Reproducibility Crisis）」的深度閱讀理解、摘要歸納與精準英文申論能力。
本題的核心考點包含：
- 名詞定義與量化數據對應（Definition and Quantitative Evidence）：精準理解 Reproducibility 的學術定義，並提取特定調查母體與百分比。
- 邏輯矛盾分析（Paradox / Contradiction Analysis）：辨識研究者在「自身遭遇的複製失敗事實」與「對已發表文獻的高度信任／對失敗指標的寬容」之間的認知落差（Cognitive Dissonance）。
- 因果關係與負面衝擊推論（Causal Factors and Impact Analysis）：辨識造成重複性困難的障礙（如出版偏差、選擇性報導、溝通缺乏等），並延伸說明其對科研誠信與資源配置的實質危害。
- 領域差異對比（Disciplinary Disparity）：橫向對比醫學研究（Medical research）與物理／工程（Physics and engineering）在採行改善措施上的比例差距。
- 解決策略彙整（Actionable Solutions）：條列機構所建議的改善方案（實驗設計、數據分析、研究培訓、標準化等）。
解題方法與作答策略
題幹明確規定「以英文簡答（Answer in brief English）」且特別註記「ATTN: Use specific examples and statistics from the passage to support your answers.」。因此，作答必須遵循以下三大原則：
- 精準點題，切勿抄整句（Paraphrasing & Synthesis）：避免不加消化地整句照抄，應重組句構、凸顯關鍵名詞與動詞，展現學術英文整合能力。
- 數據嚴謹引用（Exact Numerical Evidence）：每一題問及數據處，必須精準帶入百分比（如 1,500, >70%, >50%, 73%, 31%, 41%, 24%, 30%, 80%），不可遺漏或混淆。
- 分點作答、架構清晰：每一子題均給予清晰的主旨句（Topic Sentence），緊接著佐證數據（Supporting Data）與簡短說明。
各小題逐題解析與擬答（含中文解析）
第 1 小題
- 解析：
題目包含兩問：(1) 何謂「reproducibility」？(2) 有多少比例／數量的受訪科學家遭遇此問題？
由首句可知，reproducibility 是「research validation 的基石（a cornerstone of research validation）」，亦指重新進行實驗以獲得一致結果（methodological consistency / replicating previous work）的能力。
數據方面，調查樣本約為 1,500 位科學家（surveying approximately 1,500 scientists）；其中超過 70%（more than 70%）無法複製同儕的實驗，超過 50%（over 50%）甚至無法複製自己過去的研究成果。 - 英文參考解答：
Reproducibility refers to the ability to duplicate an experiment and achieve consistent results, serving as a fundamental cornerstone of research validation and methodological consistency. In a survey of approximately 1,500 scientists, reproducibility problems were widespread: more than 70% of researchers failed to reproduce experiments conducted by their peers, and over 50% struggled to replicate their own prior work.
第 2 小題
- 解析：
題目要求利用文章數據，解釋科學家對「reproducibility」概念所展現出的矛盾態度（contradiction）。
文章指出：- 事實上絕大多數科學家都曾經歷過複製失敗（同儕 >70%、自己 >50%），但令人意外的是（Interestingly），多數研究者仍對既有文獻高度信任——73% 的人相信自己領域內至少半數的論文是可信的（trusted）。
- 只有 31% 的科學家將「複製失敗」視為「結果無效（invalid results）」的指標。
這種「明明實驗極度難以重現，卻依然高度盲信已發表論文，且不願將重現失敗歸咎於結果無效」的現象，即為顯著的認知矛盾。
- 英文參考解答：
The contradiction lies in the disparity between the widespread failure of experimental replication and the high level of confidence scientists still place in published literature. Although over 70% of researchers failed to reproduce peers' experiments and over 50% failed with their own, 73% of respondents continued to believe that at least half of the published papers in their field are trustworthy. Furthermore, only 31% viewed a failed replication as an indicator of invalid results, demonstrating an unjustified optimism despite pervasive empirical failure.
第 3 小題
- 解析：
題目要求列舉出改善可重複性的至少兩項障礙（barriers），並說明其對科研的潛在影響（potential impact）。
文章提及的障礙包括：出版偏差與選擇性報導（publication bias and selective reporting，最顯著）、實驗測試不足（insufficient testing）、同儕審查不周（inadequate peer review）、技術困難（technical difficulties）、溝通不良（communication barrier，少於 20% 的研究者曾因重現失敗而被他人聯繫）。
考生應任選兩項，並說明影響：- 出版偏差／選擇性報導：導致學術文獻充斥陽性結果假象，誤導後續研究方向與經費資源。
- 缺乏溝通：阻礙錯誤更正與技術交流，造成重複走冤枉路與研究資源浪費。
- 英文參考解答：
Two major barriers identified in the passage are:- Publication bias and selective reporting: Scientists tend to publish only positive or successful results. Its potential impact is the distortion of the scientific record, creating a false perception of efficacy and misleading future research initiatives.
- Poor communication: Less than 20% of researchers were ever contacted by peers regarding replication failures. This lack of dialogue prevents constructive troubleshooting, hampers error correction, and leads to wasted academic resources and redundant experimental attempts.
第 2 題
請將本段落翻譯為英文(30%):
電子郵件是圖書館執行館務營運與服務讀者的重要資通訊工具,但同時也是資安威脅的主要來源。攻擊者可
能會偽造資料庫廠商的系統更新通知、期刊出版社的電子期刊訂閱確認信,或是文獻傳遞申請等釣魚郵件。
這些經過精心設計的社交工程攻擊,往往誘使使用者點選惡意連結,導致其在偽造的身份驗證頁面上輸入帳
號與密碼。一旦擁有管理者權限的帳號遭到盜用,攻擊者不僅能夠存取圖書館資訊系統的後端資料庫,更可
能藉由資料庫管理權限,竊取讀者個資與電子資源採購合約等重要資訊。
登入後即可作答並保存紀錄。
核心觀念
本題考查「科技與資安主題段落翻譯」，重點包括：
- 正確處理長句的主從關係與轉折。
- 掌握圖書館、電子資源與資安領域的專業詞彙。
- 分辨「偽造通知」與「釣魚郵件」的語意層次。
- 運用被動語態、關係子句與不定詞片語，使英文符合正式書面語。
- 清楚呈現攻擊流程：偽造郵件 → 誘導點擊 → 竊取帳密 → 濫用管理權限 → 存取資料與竊取機密資訊。
本題不是逐字替換，而是將原文的資訊層次與因果關係，以自然、正式的英文重新組織。
解題方法
翻譯時可依照下列四個步驟處理：
- 先找出每句的主幹與邏輯關係。
- 將專業名詞固定翻譯，避免同一概念前後不一致。
- 中文中較長的連續動作，改用英文的分詞片語、關係子句或不定詞片語連接。
- 完成後檢查時態、語態、單複數與搭配詞。
第一層：電子郵件的雙重角色
「是重要工具，但也是威脅來源」具有明顯轉折，可使用：
not only ... but also ...an important ... tool, but also a major source of ...
本句的「資通訊工具」可譯為 information and communications technology tool，但在英文正式語境中，直接使用 communication and information tool 不自然。此處以 information and communication tool 搭配電子郵件的功能即可。
「執行館務營運與服務讀者」可處理為：
carrying out library operationsproviding services to library users
第二層：列舉釣魚郵件的偽裝形式
「偽造……通知、……確認信，或是……」可使用：
forge phishing emails that appear to be ...
這裡的重點不是攻擊者真的偽造資料庫或出版社，而是偽造「看似來自這些機構的郵件」。因此使用：
emails purporting to be system update notifications from database vendors
比直接寫成 fake database vendors' notifications 更精確。
專業詞彙可譯為：
- 資料庫廠商：
database vendors - 系統更新通知：
system update notifications - 期刊出版社：
journal publishers - 電子期刊訂閱確認信：
electronic-journal subscription confirmation emails - 文獻傳遞申請：
document-delivery requests - 釣魚郵件：
phishing emails
第三層：社交工程攻擊的作用方式
「經過精心設計的」可用過去分詞片語：
carefully crafted social-engineering attacks
「誘使使用者點選惡意連結」可用：
trick users into clicking malicious links
trick someone into + 動名詞 是常見句型，表示「誘騙某人做某事」。
「導致其在偽造的身份驗證頁面上輸入帳號與密碼」可用現在分詞片語銜接：
leading them to enter their usernames and passwords on counterfeit authentication pages
其中：
lead someone to + 原形動詞表示導致某人做某事。counterfeit authentication pages表示偽造的身分驗證頁面。- 「帳號與密碼」可譯為
usernames and passwords，比籠統的account and password更自然。
第四層：管理者帳號遭盜用後的後果
「一旦……遭到盜用」可用條件子句：
Once an account with administrative privileges is compromised
「擁有管理者權限的帳號」以關係概念表達為：
an account with administrative privileges
「不僅……更可能……」使用：
not only ... but also ...
「存取圖書館資訊系統的後端資料庫」可譯為：
access the backend databases of the library's information systems
「藉由資料庫管理權限」可譯為：
by exploiting database-administration privileges
exploit 在資安文章中表示「利用弱點、權限或系統機制」，比一般的 use 更準確。
「讀者個資」可譯為：
library users' personal information
「電子資源採購合約」可譯為：
procurement contracts for electronic resources
參考譯文
Email is an important information and communication tool for carrying out library operations and providing services to library users, but it is also a major source of cybersecurity threats. Attackers may forge phishing emails that appear to be system update notifications from database vendors, electronic-journal subscription confirmation emails from journal publishers, or document-delivery requests. These carefully crafted social-engineering attacks often trick users into clicking malicious links, leading them to enter their usernames and passwords on counterfeit authentication pages. Once an account with administrative privileges is compromised, attackers can not only access the backend databases of the library’s information systems but also exploit database-administration privileges to steal library users’ personal information and important materials such as procurement contracts for electronic resources.
句型與用字解析
第 3 題
請將本段落翻譯為中文(30%):
The evolution of information discovery in academic libraries has been transformed by artificial intelligence
technologies. Conventional search methods, such as keyword searches and database queries, often leave researchers
overwhelmed by the vast amount of available information, making it difficult to locate relevant resources efficiently.
AI-powered solutions now offer enhanced search capabilities, including intelligent metadata extraction, automated
literature review, and personalized content recommendations. These tools not only help researchers navigate through
extensive databases more effectively but also assist librarians in improving content organization and accessibility.
While Al brings significant advantages to information discovery, libraries must carefully evaluate these tools to ensure
they maintain the quality and reliability of academic resources. The challenge lies in balancing advanced technology
with core library principles of providing authoritative and well-curated information to support scholarly research.
登入後即可作答並保存紀錄。
核心觀念
本題考查學術圖書館與人工智慧相關詞彙的理解，以及長句的中文轉譯。翻譯時須掌握段落的論述脈絡：人工智慧提升資訊探索效率，同時也帶來確保學術資源品質與可靠性的課題。
解題方法
先依語意將段落分成「傳統搜尋的困難」、「人工智慧帶來的功能與效益」及「圖書館須面對的挑戰」三部分，再辨認句子的主幹與修飾關係，最後調整語序，使中文表達自然流暢。
例如，第二句的主幹是「傳統搜尋方式常使研究者不知所措」；後面的 making it difficult... 說明其結果，宜譯為「因而難以有效率地找出相關資源」。末句的 to support scholarly research 則修飾「提供資訊」的目的，應譯為「以支持學術研究」。
重要詞語與句型
- information discovery：資訊探索，指尋找與發現所需資訊的過程。
- metadata extraction：中繼資料擷取；中繼資料是描述資源內容、來源或特徵的資料。
- literature review：文獻回顧。
- accessibility：可及性，本文指資訊容易被找到、取得與使用。