📚 AS Cambridge Computer Science: Exam Techniques and Mark Schemes | AS剑桥计算机科学：答题技巧与评分标准
Mastering the AS Cambridge Computer Science 9618 exam is not just about knowing your theory; it is about understanding exactly what examiners expect and how to present your answers effectively. This guide breaks down the exam structure, command words, mark schemes, and proven strategies for each question type. Whether you are tackling pseudocode, SQL, or long-answer theory, these techniques will help you maximise every mark.
要在 AS 剑桥计算机科学 9618 考试中取得高分，仅仅掌握理论知识是不够的；你还必须理解考官究竟想要什么，以及如何有效地呈现你的答案。本篇指南将详细拆解考试结构、命令词、评分方案，并针对每种题型提供经过验证的答题策略。不论你面对的是伪代码、SQL 还是长篇论述题，这些技巧都能帮助你争取到每一分。
1. Understanding the Exam Structure and Mark Schemes | 理解考试结构与评分方案
The AS Computer Science syllabus (9618) is assessed through two papers. Paper 1 Theory Fundamentals lasts 1 hour 30 minutes and carries 75 marks, focusing on information representation, communication, hardware, processor fundamentals, system software, security, ethics, and databases. Paper 2 Fundamental Problem-solving and Programming Skills is 2 hours long, also 75 marks, and includes a scenario-based section that uses pre-release material issued about six weeks before the exam, plus questions on algorithm design, programming, and testing.
AS 计算机科学 (9618) 通过两套试卷来考核。卷一理论基础知识时长 1 小时 30 分钟，满分 75 分，侧重信息表示、通信、硬件、处理器基础、系统软件、安全、伦理及数据库。卷二基础问题解决与编程技巧时长 2 小时，同样 75 分，其中包含一个基于场景的部分，会用到考前约六周发放的预发布材料，此外还有算法设计、编程和测试题。
Mark schemes reward correct knowledge with B marks, the application and interpretation of knowledge with C marks, and evaluation or higher-order skills with A marks. B marks are for recalling definitions and facts; C marks are for applying concepts to unfamiliar contexts; A marks are for analysis, justification, and design decisions. Learn to identify which type each sub-question demands, because writing a description when an evaluation is required will lose marks.
评分方案用 B 分数奖励正确的知识，用 C 分数评测知识的应用与解释，用 A 分数考查评价或高阶技能。B 分数针对定义和事实的记忆；C 分数要求将概念应用到陌生情境中；A 分数则涉及分析、论证与设计决策。你必须学会判断每个小问究竟要考查哪一类能力，如果题目要求评价而你只写了描述，就会丢分。
2. Using Pre-release Material Strategically | 战略运用预发布材料
Paper 2 includes a scenario based on the pre-release material. You receive this document about six weeks before the exam, and you must study it carefully. Do not simply read it once; annotate it, identify the core problem to be solved, list any constraints, inputs, processes, and outputs. Build a glossary of key terms used in the scenario, and practise writing algorithms that address part of the stated problem.
卷二含有一个基于预发布材料的场景题。这份文档大约在考前六周发给你，你必须认真钻研。不要只读一遍了事；要加以批注，找出需要解决的核心问题，列出所有的约束条件、输入、处理过程和输出。为材料中使用的关键术语建立一份词汇表，并练习编写能够解决其中部分问题的算法。
Examiners often extend the scenario in the exam paper by introducing a new requirement or altering a constraint. Prepare by asking yourself 'what if' questions. For example, if the pre-release describes a booking system, consider how you would modify the design for peak-time discounts or error handling. Having a skeleton plan for several variations will save precious time and let you quickly adapt your algorithms in the exam.
考官常常会在试题中延伸该场景，比如引入一项新需求或改变某个约束条件。通过问自己“如果……会怎样”来做好准备。例如，如果预发布材料描述了一个预约系统，那就思考如何修改设计以适应高峰时段折扣或错误处理。为几种变化准备好骨架方案，就能在考试中节省宝贵时间，并快速调整你的算法。
3. Command Words and How to Address Them | 命令词及其应对
Command words such as 'State', 'Describe', 'Explain', 'Compare', 'Evaluate', and 'Discuss' determine the depth and style of your answer. 'State' requires a short, factual response, often just one word or a phrase. 'Describe' asks for an account of what something is or does, without justification. 'Explain' requires a reason or cause, often using 'because'. 'Compare' needs similarities and differences. 'Evaluate' demands you weigh up advantages and disadvantages and come to a reasoned conclusion. 'Discuss' involves presenting both sides of an issue and offering a balanced judgement.
诸如 "State"（陈述）、"Describe"（描述）、"Explain"（解释）、"Compare"（比较）、"Evaluate"（评价）和 "Discuss"（讨论）这类命令词决定了答案的深度与风格。 "State" 要求简短、事实性的回答，通常是一个词或短语。 "Describe" 要求说明某事物是什么或做什么，无需理由。 "Explain" 则需要给出原因或缘由，往往要用到“因为”。 "Compare" 需要找出相似点和不同点。 "Evaluate" 要求权衡利弊并得出有理有据的结论。 "Discuss" 则涉及呈现问题的两方面并给出平衡的判断。
Never ignore the command word. If a question says 'Evaluate the use of parity checks', but you only describe how parity works, you will miss the A marks entirely. Underline the command word and structure your answer accordingly. A good evaluation might state, 'Parity checks can detect single-bit errors, but they cannot identify which bit is corrupted and cannot correct errors, making them less suitable for critical data.'
永远不要忽视命令词。如果题目要求 "评价奇偶校验的使用"，而你却只描述了奇偶校验如何工作，你将完全丢掉 A 分数。在命令词下画线，并据此组织你的答案。一段好的评价可以这样写："奇偶校验能检测单比特错误，但它无法识别出哪一个比特出错，也不能纠正错误，因此不适合用于关键数据。"
4. Theory Questions: Recall and Explanation | 理论题：记忆与解释
Paper 1 is dominated by theory questions that test your knowledge of computer systems. Use precise technical vocabulary. For instance, instead of saying 'memory loses data when turned off', write 'volatile memory loses its contents when the power supply is removed'. Write definitions exactly as they appear in the syllabus or reputable textbooks, as approximations may not gain the B mark.
卷一以理论题为主，测试你对计算机系统的认知。请使用精确的技术词汇。例如，不要说 "内存在断电时会丢失数据"，而应写为 "易失性存储器在电源被切断时会丢失其内容"。严格按照大纲或权威教科书中的原文来撰写定义，近似说法可能无法拿到 B 分数。
Many students lose marks by writing too little or too much. A three-mark 'Explain' question needs three distinct points, each with a cause or effect. Practise timing: for a 75-mark paper in 90 minutes, you have roughly 1.2 minutes per mark. Use this to decide how much to write. Bullet points are often acceptable and can keep your answer clear, but ensure each bullet is a complete, relevant point.
很多学生因为写得过少或过多而丢分。一道 3 分的解释题需要三个不同的要点，每个要点都要有原因或影响。练习时间控制：90 分钟的试卷 75 分，大约每分对应 1.2 分钟。据此决定写多少。使用项目符号通常是可以的，并能让答案更加清晰，但要保证每条都是一个完整且相关的点。
5. Programming and Pseudocode Excellence | 编程与伪代码的精通
For programming questions, the examiner expects a clear, logical solution. You may use Python, Java, C#, or the Cambridge pseudocode. Whichever you choose, stick to it consistently. Pseudocode is strongly recommended because it avoids syntax-heavy penalties and focuses on algorithmic thinking. Ensure your pseudocode follows the syllabus style: use keywords like INPUT, OUTPUT, IF … THEN … ELSE … ENDIF, WHILE … DO … ENDWHILE, and FOR … TO … NEXT.
在编程题中，考官期待清晰、逻辑严谨的解答。你可以使用 Python、Java、C# 或剑桥伪代码。无论选用哪一种，都要保持前后一致。强烈推荐使用伪代码，因为它可以避免因严格语法而扣分，并聚焦于算法思维。确保你的伪代码符合大纲风格：使用 INPUT、OUTPUT、IF … THEN … ELSE … ENDIF、WHILE … DO … ENDWHILE 以及 FOR … TO … NEXT 等关键字。
Always declare variables with a clear comment or data type hint. Use meaningful identifiers such as TotalCost instead of x. Remember to initialise counters and accumulators. Include final output statements to print results. Examiners will trace your logic mentally, so indentation and structure are vital. Even if the algorithm is not perfect, showing a planned approach with comments can attract partial marks under 'compensation error' rules.
始终通过清晰的注释或数据类型提示来声明变量。使用有意义的标识符如 TotalCost 而不是 x。务必初始化计数器和累加器。加上最终的输出语句来打印结果。考官会在头脑中跟踪你的逻辑，因此缩进和结构至关重要。即使算法不完美，用注释展示规划过程也能根据“补偿错误”规则获得部分分数。
6. Flowcharts, Trace Tables and Dry Runs | 流程图、跟踪表与走查
Flowchart questions test your ability to represent algorithms graphically. Use standard symbols: ovals for start/end, parallelograms for input/output, rectangles for processes, diamonds for decisions. All flow lines should have arrows showing the direction of flow. If you are asked to convert pseudocode to a flowchart, make sure every statement is represented.
流程图题目考查的是用图形表示算法的能力。使用标准符号：椭圆表示开始/结束，平行四边形表示输入/输出，矩形表示处理过程，菱形表示判断。所有流程线都应带有箭头以指明流向。如果题目要求将伪代码转换为流程图，务必确保每个语句都有对应的图形符号。
Trace tables are a powerful tool to check your algorithm's correctness. When completing a trace table in the exam, list all variables in the header row and add a new row each time any variable changes value. Use a 'print' column if there is output. Practice dry running your own code with trace tables at home; this sharpens your ability to spot logical errors and will help you answer 'complete the trace table' questions quickly and accurately.
跟踪表是检查算法正确性的强大工具。在考试中填写跟踪表时，在表头行列出所有变量，每次任一变量值发生变化时就新增一行。如果有输出，则增加“打印”列。在家练习时用跟踪表对自己编写的代码进行走查，这能提高发现逻辑错误的能力，并帮助你快速且准确地回答“完成跟踪表”类题目。
7. Algorithm Design and Evaluation | 算法设计与评估
When asked to design an algorithm for a scenario, start by identifying inputs, processes, and outputs. Jot these down as comments before writing any code. Break the problem into sub-problems using a top-down approach. For example, a student grading system might have sub-problems: input marks, validate range, calculate average, determine grade, and output result.
当题目要求为某个场景设计算法时，首先要确定输入、处理过程和输出。在开始编写代码之前，先用注释的形式将这些记下来。运用自顶向下的方法将问题分解为子问题。例如，学生成绩系统可分解为：输入分数、验证范围、计算平均分、评定等级和输出结果等子问题。
Evaluation questions often follow a design task. Be prepared to discuss efficiency (e.g. number of loops, use of appropriate data structures), readability (meaningful names, comments), and robustness (validation checks, error handling). You might be asked to suggest improvements. Always refer back to the given scenario and explain why a change is better. For instance, 'Using a FOR loop instead of a WHILE loop improves readability because the number of iterations is known.'
评价题往往紧跟在设计任务之后。准备好从效率（如循环次数、是否使用了合适的数据结构）、可读性（有意义的命名、注释）和健壮性（验证检查、错误处理）等方面进行讨论。你可能需要提出改进建议。务必结合给定的场景，并解释为什么修改更好。例如，“使用 FOR 循环代替 WHILE 循环可以提高可读性，因为迭代次数是已知的。”
8. Database and SQL Solutions | 数据库与SQL解答
Database questions typically ask you to write SQL queries, understand flat-file vs relational databases, normalise tables, or draw entity-relationship diagrams. When writing SQL, use correct syntax: SELECT column1, column2 FROM table WHERE condition;. Pay attention to sorting (ORDER BY), grouping (GROUP BY) and aggregate functions (COUNT, SUM, AVG). Always end SQL statements with a semicolon.
数据库题目通常要求你编写 SQL 查询，理解平面文件与关系数据库的区别，规范化表格，或者绘制实体关系图。编写 SQL 时请使用正确语法：SELECT column1, column2 FROM table WHERE condition; 。注意排序(ORDER BY)、分组(GROUP BY)以及聚合函数(COUNT, SUM, AVG)的使用。SQL 语句末尾务必加上分号。
If a question asks you to provide a query for 'all customers who have placed more than three orders', remember to use GROUP BY … HAVING COUNT(*) > 3. A common mistake is to use WHERE with an aggregate function, which is incorrect. Normalising tables to Third Normal Form (3NF) requires removing partial and transitive dependencies. Present your normalised tables clearly, denoting primary and foreign keys with PK and FK.
如果题目要求你给出“所有订单数超过三个的客户”的查询，记得使用 GROUP BY … HAVING COUNT(*) > 3。常见的错误是在聚合函数上使用 WHERE，这是不正确的。将表格规范化到第三范式(3NF)需要消除部分依赖和传递依赖。清晰地呈现规范化后的表格，用 PK 和 FK 分别标注主键和外键。
9. Computer Architecture and Hardware | 计算机体系结构与硬件
Topics such as the fetch-decode-execute cycle, von Neumann architecture, memory types, and assembly language require precision. Diagrams can earn marks, but they must be accompanied by labels and a brief explanation. When describing the fetch phase, mention the program counter (PC), memory address register (MAR), memory data register (MDR), and current instruction register (CIR).
抓取-解码-执行循环、冯·诺依曼体系结构、存储器类型和汇编语言等主题要求答题高度精确。图表可以得分，但必须加上标签和简要说明。在描述抓取阶段时，要提及程序计数器(PC)、内存地址寄存器(MAR)、内存数据寄存器(MDR)和当前指令寄存器(CIR)。
Understand the roles of different buses (address, data, control) and how they interact. When comparing RISC and CISC, provide more than one distinction: RISC uses simple, single-cycle instructions and requires a larger number of instructions, while CISC uses complex, multi-cycle instructions but fewer of them. Always link architecture features to their effect on performance, such as pipelining and cache memory.
要理解不同总线（地址总线、数据总线、控制总线）的作用以及它们如何交互。在比较 RISC 与 CISC 时，要提供不止一个区别：RISC 使用简单的单周期指令，指令数量较多；而 CISC 使用复杂的多周期指令，但指令数量较少。要始终将体系结构的特征与其对性能的影响联系起来，例如流水线技术和高速缓存。
10. Networking, Communication and Ethics | 网络、通信与伦理
Questions on networking cover topologies, protocols, layers, and error detection. Use the OSI or TCP/IP model to structure explanations. For example, when asked how an email is sent, trace through the layers: application layer (SMTP), transport layer (TCP), network layer (IP) and data link/physical layers. name specific protocols and their functions.
涉及网络的题目涵盖拓扑结构、协议、分层和错误检测。利用 OSI 或 TCP/IP 模型来组织你的解释。例如，当被问到电子邮件如何发送时，可逐层追溯：应用层(SMTP)、传输层(TCP)、网络层(IP)以及数据链路层/物理层。说出具体协议名称及其功能。
Ethics, law, and data security questions often require you to discuss impacts. Avoid generic statements like 'It is unsafe'. Instead, specify threats (malware, phishing, SQL injection) and corresponding countermeasures (firewalls, encryption, access rights). For ethical debates, present both the advantages of a technology (e.g., faster diagnosis in medical AI) and its drawbacks (privacy concerns, bias), and then give a reasoned conclusion. Cambridge mark schemes reward balance and depth.
伦理、法律与数据安全题目通常要求你展开讨论。避免诸如“这很不安全”之类的笼统说法。要具体说明威胁（恶意软件、网络钓鱼、SQL 注入）以及相应的对策（防火墙、加密、访问权限）。对于伦理讨论，要同时呈现技术的优点（如医疗 AI 能更快诊断）和缺点（隐私担忧、偏见），然后给出有理有据的结论。剑桥的评分方案奖励平衡且深入的论述。
11. Time Management and Paper Technique | 时间管理与应试技巧
Start by scanning the entire paper to assess the questions. Tackle the questions you feel most confident about first, but do not spend more than the allocated minutes per mark. For a 6-mark question, aim to finish in under 8 minutes. If you get stuck, make a brief note and move on; you can return later with fresh eyes.
先快速浏览整张试卷，评估各题的难度。从你最有把握的题目开始做，但每题花费的时间不要超过按分值分配的时间。一道 6 分的题目要争取在 8 分钟内完成。如果卡住了，就做个简短标记并跳过，回头再看时思路会更清晰。
For programming questions on Paper 2, plan your algorithm for 2-3 minutes before coding. Write a few pseudo-comments outlining the structure. This prevents you from going down a dead-end path that wastes time. Leave the last 5-10 minutes to review your answers. Check for missing semicolons in SQL, uninitialised variables in pseudocode, and that every command word has been addressed appropriately.
对于卷二的编程题，先花 2 到 3 分钟规划算法再开始写代码。用伪注释勾勒出整体结构。这能防止你走上一条耗费时间的死胡同。留出最后 5 到 10 分钟检查答案。检查 SQL 中是否漏掉分号，伪代码中变量是否已初始化，以及每个命令词是否都得到了恰当的回应。
12. Common Mistakes and Self-Assessment Using Mark Schemes | 常见错误与评分方案自评
Frequent mistakes include: confusing B and A mark demands, writing pseudocode that is too language-specific (e.g., using Python's print() syntax instead of OUTPUT), not reading the pre-release material thoroughly, and failing to show working in trace tables. Another major pitfall is neglecting the 'Explain' part of an evaluation, providing only a list of features without reasoning.
常见错误包括：混淆 B 分数与 A 分数的要求，编写的伪代码过于针对具体语言（例如使用 Python 的 print() 语法而非 OUTPUT），没有仔细阅读预发布材料，以及在跟踪表中未展示推导过程。另一个大坑是忽略了“评价”中的“解释”部分，只罗列特征而不提供原因。
The official mark schemes are your best revision tool. Download past papers and mark schemes from the Cambridge website. After attempting a question, compare your answer line by line with the scheme. Note what the examiner accepts as alternative correct answers. Pay attention to the mark allocation in brackets, e.g. (3 marks) because it tells you the number of distinct points expected. Maintain a 'common errors' log and review it before the exam.
官方评分方案是你最好的复习工具。从剑桥网站下载历年真题与评分方案。尝试回答一道题后，逐行与方案对照。留意考官接受哪些替代正确答案。注意方括号中的分值分配，如 (3 分)，因为它告诉了你应给出多少个不同的要点。建立一份“常见错误”日志，并在考前复习。
Published by TutorHao | Computer Science Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图