本课为原创大学书面英语能力课程，不对应统一的人教版大学教材或固定大学年级；不含听力。练习目标只反映本站材料掌握情况，不保证实际考试分数。
学习目标
评估算法推荐主题中的目标、指标和用户选择，形成有边界的书面回应。
准确完成同主题摘要、汉译英和论证写作，避免抄句式替代思考。
在本课主题中理解新增词块的语境用法，并把其中相关词项迁移到新的完整句子。
核心讲解
综合能力要求读懂“系统优化什么”和“使用者需要什么”之间的差异。recommendation在科技语境指推荐内容，不是老师写的推荐信；engagement可能指参与/互动指标，不自动等于获益；personalised不意味着完全理解每个人。
原创阅读：A recommendation is a starting point
A learning website recommended articles based on what users had clicked before. Its designers wanted to help readers find material quickly. During a trial, recommended articles received more clicks than articles chosen at random from the same collection. The team regarded this as evidence that previous choices could help predict what would attract attention.
A teacher who used the site welcomed the convenience but raised another question. If a student repeatedly clicked short texts on familiar subjects, would the system keep offering the same kind of material? Interest could make reading easier to begin, but learning sometimes required an unfamiliar topic or a more demanding argument. A recommendation could be attractive without being the most useful next step.
The designers added an explanation beside each suggestion and a separate option labelled “Try something different”. Users could also adjust the level they wanted to practise. These changes made the system's assumptions more visible, but they did not prove that learning had improved. The team would need evidence about what users understood and could later do, not merely what they selected.
The teacher did not argue that recommendations should be removed. Instead, she treated them as starting points for choice. A useful reading plan might combine familiar material that builds confidence with less familiar texts that broaden understanding. Keeping the purpose of learning explicit mattered more than making every selection automatic.
四级阶段能抓“增加点击不等于提高学习”；提高阶段进一步解释指标、目标、可选择性与适当挑战之间的关系，并提出可评估的改进。
例题解析
不准确推断：The system prevented every student from learning anything new.
原文只是提出重复熟悉主题的风险；后面增加探索选项，没有全体学习结果。应写：A system based only on past clicks may favour familiar material and therefore may not always support broader learning. 两处may不是拖泥带水，而是保留证据强度。
本课词汇与搭配
下列为本课实际讲解的词项；词形变化及同一词的不同义项不等于新增独立词汇量。
| 词项 | 本课义项及用法 |
|---|---|
| recommendation | n. 推荐；a reading recommendation |
| personalised | adj. 个性化的；personalised content |
| engagement | n. 参与／互动；user engagement |
| predict | v. 预测；predict a choice |
| demanding | adj. 要求高的；a demanding argument |
| broaden | v. 拓宽；broaden understanding |
| explicit | adj. 明确的；an explicit goal |
| assumption | n. 假设；make assumptions visible |
| selection | n. 选择；the basis for selection |
词汇工坊
本表补足本课话题的读写词块，含中学承接和大学应用，不将每项都计为大学新词。先遮住中文解释推测例句，再用自己的句子替换人物、时间或对象。理解义项、掌握搭配、能够迁移是三项不同任务；下方两道迁移练习检验部分关键词块，不代表全部词项已自动掌握。
| 词项 | 义项与搭配 | 原创语境例句 |
|---|---|---|
| personalisation | 个性化；content personalisation | Content personalisation can use past choices without understanding every goal. |
| recommendation system | 推荐系统；evaluate a recommendation system | Evaluate a recommendation system against the learner's purpose. |
| click-through | 点击行为；click-through measures | Click-through measures show selection, not necessarily understanding. |
| exposure | 接触；exposure to unfamiliar topics | Exposure to unfamiliar topics may broaden a reading plan. |
| filter bubble | 信息过滤圈；a possible filter bubble | The discussion raised a possible filter bubble rather than proving one for every user. |
| curation | 筛选整理；human curation | Human curation can complement automated suggestions. |
| novelty | 新颖性；balance novelty and familiarity | A reading plan can balance novelty and familiarity. |
| familiarity | 熟悉度；subject familiarity | Subject familiarity may make an article easier to begin. |
| transfer | 迁移；transfer learning to a new task | A new task helps test whether learning transfers beyond the original text. |
| retention | 保留记忆；check retention | A delayed summary can help check retention. |
| metacognition | 元认知；metacognitive awareness | Metacognitive awareness includes noticing what we do not yet understand. |
| self-regulation | 自我调节；support self-regulation | A clear goal can support self-regulation during study. |
| calibration | 校准；confidence calibration | Confidence calibration compares how sure a learner feels with actual performance. |
| learning objective | 学习目标；a measurable learning objective | A measurable learning objective describes what the learner should be able to do. |
练习
先独立完成，再核对答案；开放题须按要点人工评分，不以是否逐字等同范文评分。
1
试验直接显示了什么？又没有证明什么？
2
教师为什么不把更高点击率等同于更好学习？
3
用40—60词英文总结原文，包含建议用途与限制。
4
译：“个性化推荐可以帮助读者快速找到感兴趣的文章，但过去的选择不应成为唯一依据。平台应让用户了解推荐原因，并保留探索不同主题的机会。”
5
写160—200词短文：How should students use reading recommendations? 既讨论便利，也讨论选择局限，提出两项具体做法。
6
给写作设计一种比“点开多少篇”更贴近理解的检查方法，并说明它仍有什么限制。
7
词汇迁移：译：推荐系统应平衡新颖性与熟悉度，而不只是优化点击指标。
8
词汇迁移：译：延迟摘要可以检查记忆保留，而新任务可以测试学习是否迁移到了陌生情境。
参考答案与解析 · 完成练习后展开
参考答案与解析
1
直接显示在这次试验中推荐文获得更多点击；未证明学习理解和后续运用能力改善。没有进一步数据就不能写成绩提升。
2
点击可能反映熟悉、短、容易吸引注意；学习有时需要陌生主题和更具挑战的论证。兴趣是起点，不是全部学习效果。
3
The website used past clicks to recommend reading, which attracted attention but did not establish better learning. A teacher suggested keeping recommendations as starting points rather than automatic decisions. Explanations, adjustable levels and opportunities to try unfamiliar topics could support choice, while actual understanding would still need to be assessed. 10分：机制2、限制3、回应3、语言2。
4
Personalised recommendations can help readers quickly find articles that interest them, but past choices should not be the only basis for selection. Platforms should explain why articles are recommended and preserve opportunities for users to explore different topics. 15分：便利3、非唯一依据4、解释原因3、探索机会3、语言2。可用customised suggestions，但personalised recommendations更贴合科技语域。
5
示例：Students should treat reading recommendations as useful suggestions rather than complete learning plans. A recommended article can reduce the time spent searching and make it easier to begin reading, particularly when the subject already interests the learner.
However, convenience is not the only goal. A system that relies on past choices may repeatedly suggest familiar topics or easy texts. Students could miss opportunities to examine a different viewpoint or practise a more demanding structure. They should therefore ask why an article was recommended and whether it matches a current learning goal.
Two habits can help. First, students can combine recommended texts with material from topics they rarely choose. Second, after reading, they can write a brief summary and explain one piece of evidence without looking at the article. This checks more than whether a title attracted attention.
Recommendations do not need to disappear. They become more valuable when students remain responsible for choosing, checking their understanding and gradually widening the range of material they can read.
20分：观点与任务4、便利和限制6、两做法4、组织2、语言4。
6
可让学习者隔日不看原文写主旨与一条证据，再用新例子说明同一关系。局限：摘要评分需人工标准，且单次任务不能代表全部能力。8分：贴合理解4、可执行2、限制2。
7
A recommendation system should balance novelty and familiarity rather than optimise click-through measures alone. balance A and B；alone限制“只看点击”。6分：信息3，词块2，句法1。
8
A delayed summary can check retention, while a new task can test whether learning transfers to an unfamiliar setting. while对比两种用途，不表示因果。6分：信息3，搭配2，句法1。