📚 High-Frequency Topics and Common Pitfalls in Pre-U WJEC Computer Science | Pre-U WJEC 计算机：高频考点与易错题分析
The WJEC Pre-U Computer Science qualification is known for its rigorous blend of computational thinking, programming mastery, and deep theoretical understanding. Candidates often excel in isolated areas yet stumble on cross-topic questions that demand synthesis and precision. This revision guide identifies the most frequently examined themes, unpacks the common errors seen year after year, and provides targeted strategies to turn weaknesses into strengths.
WJEC Pre-U 计算机科学课程以其对计算思维、编程技能的扎实掌握以及深度的理论理解而著称。考生往往在独立知识点上表现出色，却在需要综合和精准作答的跨主题题目中失分。本文梳理了最高频的考点，剖析了历年反复出现的常见错误，并提供有针对性的策略，帮助你将薄弱点转化为优势。
1. Data Representation and Binary Arithmetic | 数据表示与二进制运算
Binary addition and subtraction using two's complement is tested almost every year. A typical mistake is forgetting to extend the sign bit when converting to a wider representation, which leads to overflow errors in the least expected places. For example, when adding 8-bit numbers and storing the result in a 16-bit register, failing to sign-extend can turn a positive sum into a negative value.
二进制补码的加减法几乎是每年的必考题。一个典型的错误是在转换为更宽位数的表示时忘记扩展符号位，这会导致意想不到的溢出错误。例如，将两个8位数相加并将结果存入16位寄存器时，若未进行符号扩展，可能将正数之和错误表示为负数。
Normalisation of floating point numbers in the form mantissa × 2exponent causes confusion between positive and negative mantissas. Many candidates attempt to normalise a negative mantissa by shifting left until the first two bits are '10', but they fail to adjust the exponent correctly or do not recognise that the leading bits of a normalised negative mantissa must be '10' (for a negative number) while a positive must start with '01'.
浮点数规格化（格式为尾数 × 2指数）中，正尾数和负尾数的规则常被混淆。许多考生试图将负尾数左移，直到前两位为“10”，但却未能正确调整指数，或未意识到规格化负尾数的前导位必须是“10”，而正尾数前导位则必须是“01”。
Rounding errors and precision loss in fixed-point representations also appear in analysis questions. Candidates often describe truncation but not rounding, or they confuse absolute error with relative error when comparing the accuracy of two representations.
定点数表示中的舍入误差和精度损失也常出现在分析题中。考生往往只能描述截断而非舍入，或者在比较两种表示方式的精度时，混淆了绝对误差与相对误差。
2. Boolean Algebra and Logic Simplification | 布尔代数与逻辑简化
The application of De Morgan's laws and the use of Karnaugh maps to minimise logic expressions are high-frequency skills. A recurrent error is drawing the K-map grid with an incorrect ordering of variables, causing adjacent cells not to differ by a single bit. The Gray code sequence for rows and columns must be rigorously maintained, yet under time pressure, students often revert to binary order.
德摩根定律的应用和卡诺图化简逻辑表达式是高频技能。一个反复出现的错误是绘制卡诺图网格时变量顺序错误，导致相邻单元格并非仅有一位不同。行和列的格雷码顺序必须严格保持，但在时间压力下，学生常会错误使用二进制顺序。
Another pitfall is incomplete grouping in K-maps. Although overlapping groups are allowed and often necessary to achieve a minimal sum-of-products, many candidates are reluctant to overlap, leaving isolated 1s that increase the literal count. Examiners specifically look for the minimal expression; a non-minimal solution loses marks even if functionally correct.
另一个易错点是卡诺图中的不完全分组。尽管允许且经常需要重叠分组以获得最简与或式，但许多考生不愿重叠，留下孤立的1，增加了文字数量。考官明确要求最简表达式；即使功能正确，非最简解也会失分。
When simplifying using Boolean laws, candidates often misapply the distributive or absorption laws. For instance, A + (A · B) is frequently incorrectly simplified to B instead of A. Emphasising the proof of each step can prevent such slips.
在运用布尔定律化简时，考生常误用分配律或吸收律。例如，A + (A · B) 常被错误地化简为 B 而非 A。在每一步中强调证明可避免此类疏漏。
3. Hardware Architecture and the Fetch-Decode-Execute Cycle | 硬件架构与取指-译码-执行周期
The role of registers such as the Program Counter (PC), Memory Address Register (MAR), Memory Data Register (MDR), and Current Instruction Register (CIR) must be described with precision. A common mistake is stating that the PC is incremented after the instruction is decoded, rather than after the fetch stage, or confusing the MAR's role (holding the address to be accessed) with that of the MDR (holding the data transferred).
程序计数器（PC）、存储器地址寄存器（MAR）、存储器数据寄存器（MDR）和当前指令寄存器（CIR）等寄存器的作用必须精确描述。常见错误是说PC在指令译码后而不是在取指阶段后递增，或混淆MAR（保存待访问地址）和MDR（保存传输的数据）的功能。
Interrupt handling is a topic where examiners differentiate top candidates. Many scripts describe the interrupt register and the saving of the PC to the stack but fail to mention the disabling of further interrupts during the Interrupt Service Routine or the restoration of the flag register. The sequence of checking for interrupts at the end of each fetch-decode-execute cycle must be clearly stated.
中断处理是区分优秀考生的一个主题。许多答卷描述了中断寄存器和将PC压入栈，但未提到在执行中断服务程序期间禁用其他中断，或标志寄存器的恢复。必须在每个取指-译码-执行周期末尾检查中断的顺序要阐述清楚。
Pipelining is often poorly explained in terms of data hazards. Candidates describe the concept of overlapping instruction execution but cannot identify a read-after-write hazard and the need for pipeline stalls or forwarding paths. A simple diagram tracing values through pipeline stages can earn high marks.
流水线技术中数据冒险常被解释得很差。考生描述了指令执行重叠的概念，但无法识别写后读（RAW）冒险以及流水线停顿或转发路径的必要性。通过绘制流水线各级数值传递的简图可得高分。
4. Assembly Language and Addressing Modes | 汇编语言与寻址模式
Writing small assembly code segments within the context of a given instruction set architecture (often a simplified accumulator-based or register-based model) is a staple of Pre-U papers. A persistent error is confusing immediate addressing with direct addressing. When the operand is a value prefixed with #, it must be treated as a literal; forgetting the # symbol in written code leads to accessing memory at that address instead.
在给定的指令集体系结构（通常是简化的基于累加器或基于寄存器的模型）中编写小型汇编代码片段是Pre-U试卷的常见题。一个持久性错误是将立即寻址与直接寻址相混淆。当操作数是带#前缀的值时，必须将其视为字面量；在编写的代码中忘记#符号会导致访问该地址处的内存而非值。
Indirect addressing and indexed addressing cause confusion, particularly when combined with registers. Candidates often miscalculate the effective address by adding the offset to the wrong base register. In trace-table questions, failing to update the index register as the loop progresses is a frequent slip. Examiners advise boxing the registers and their contents to track changes cycle by cycle.
间接寻址和变址寻址容易混淆，尤其当与寄存器结合时。考生常通过将偏移量加到错误的基址寄存器上来误算有效地址。在追踪表题目中，循环进行时未更新变址寄存器是常见的疏漏。考官建议将寄存器及其内容框出，逐周期追踪变化。
5. Programming Paradigms and Abstract Data Types | 编程范型与抽象数据类型
Questions on abstract data types (ADTs) such as stacks, queues, linked lists, and binary trees require both conceptual understanding and code implementation. The most common error in stack and queue operations is not handling overflow or underflow conditions. Even in pseudocode, returning a meaningful error or flag is expected.
关于栈、队列、链表和二叉树等抽象数据类型（ADT）的题目既要求概念理解，又要求代码实现。栈和队列操作中最常见的错误是未处理溢出或下溢。即使在伪代码中，返回有意义的错误或标志也是预期的。
Binary tree traversal algorithms (pre-order, in-order, post-order) are frequently mistranscribed. Candidates swap the order of visiting the left subtree, root, and right subtree for in-order, or they incorrectly apply recursion that leads to an infinite loop by failing to check for a null node. Dry runs with a simple tree are essential before attempting a recursive solution.
二叉树遍历算法（前序、中序、后序）常被错误转写。考生在写中序遍历时会把访问左子树、根节点和右子树的顺序搞混，或者错误应用递归导致未检查空节点而陷入无限循环。在尝试递归解法前，用一棵简单的树进行手工运行至关重要。
Object-oriented programming questions assess encapsulation, inheritance, and polymorphism. A pitfall is confusing method overloading (compile-time polymorphism) with method overriding (run-time polymorphism). When asked to explain the advantages of polymorphism, candidates often give generic textbook definitions without linking to code reuse and maintainability.
面向对象编程题目评估封装、继承和多态。一个易错点是混淆方法重载（编译时多态）和方法重写（运行时多态）。当被要求解释多态的优点时，考生常给出泛泛的教科书定义，而未联系到代码重用和可维护性。
6. Algorithms, Complexity, and Sorting | 算法、复杂度与排序
Big-O notation questions demand more than recalling the complexity of quicksort or merge sort. Candidates must compare algorithms for the same task and justify the choice based on best, average, and worst-case scenarios. The mistake is neglecting the space complexity of recursive algorithms, such as the O(log n) stack space for quicksort's recursion.
大O表示法题目要求的不只是回忆快排或归并排序的复杂度。考生必须对同一任务的不同算法进行比较，并根据最好、平均和最坏情况下的场景来证明选择的合理性。常犯的错误是忽略递归算法的空间复杂度，例如快速排序递归所需的 O(log n) 栈空间。
Implementations of binary search often contain off-by-one errors in the calculation of the midpoint (low + high) / 2 or in the update of the low and high boundaries. When the target is not found, many candidates forget to exit the loop, causing an infinite search. Tracing with a three-element array exposes such logic flaws.
二分查找的实现在中点计算 (low + high) / 2 或更新 low 和 high 边界时经常出现差一错误。当目标未找到时，许多考生忘记退出循环，导致无限查找。用三元素数组进行追踪能够暴露此类逻辑缺陷。
Graph traversal algorithms, particularly Dijkstra's shortest path, require careful updating of tentative distances. A common fault is updating a node's distance when a shorter path is found but failing to reorder the priority queue, or incorrectly assuming that once a node is visited its distance cannot change. The distinction between visited and optimised nodes must be explicit.
图遍历算法，尤其是Dijkstra最短路径，需要小心更新暂定距离。一个常见缺陷是当发现更短路径时更新了节点距离，却未能对优先队列重新排序，或错误地认为一旦访问过一个节点，其距离就不会再变。必须明确区分已访问节点和已优化节点。
7. Operating Systems and Resource Management | 操作系统与资源管理
Scheduling algorithms (Round Robin, Shortest Job First, Multi-level Feedback Queue) and their impact on turnaround time and waiting time are frequently examined. Candidates often miscalculate the average waiting time by forgetting that the waiting time of a process includes all the time it spends in the ready queue before completion. In Round Robin, the time quantum size drastically affects performance, yet students rarely mention the trade-off between context switch overhead and response time.
调度算法（轮转调度、最短作业优先、多级反馈队列）及其对周转时间和等待时间的影响是高频考点。考生常错误计算平均等待时间，忘了进程的等待时间包括其完成前在就绪队列中花费的所有时间。在轮转调度中，时间片大小显著影响性能，但学生很少提及上下文切换开销与响应时间之间的权衡。
Memory management concepts—paging, segmentation, and virtual memory—are confused. A typical wrong answer describes paging as dividing memory into variable-sized segments. The distinction between a page (fixed size) and a segment (logical unit) must be internalised. When explaining page faults, candidates often miss the role of the translation look-aside buffer (TLB) and the performance penalty of a TLB miss.
内存管理概念——分页、分段和虚拟内存——常被混淆。典型的错误答案是称分页将内存划分为大小可变的分段。必须深刻内化页（固定大小）和段（逻辑单元）的区别。解释缺页时，考生常忽略转换后备缓冲器（TLB）的作用以及TLB不命中的性能损失。
Deadlock detection and prevention questions require the application of the four necessary conditions—mutual exclusion, hold and wait, no preemption, and circular wait. Many candidates can list them but fail to explain how a specific prevention strategy (e.g., requiring processes to request all resources at once) breaks the hold and wait condition. A resource allocation graph must be correctly interpreted to identify a cycle as a necessary but not sufficient condition for deadlock in a multi-instance resource system.
死锁检测和预防题目要求应用四个必要条件——互斥、保持并等待、不可剥夺和循环等待。许多考生能列出它们，但未能说明具体的预防策略（例如要求进程一次性请求所有资源）如何打破保持并等待条件。必须正确解读资源分配图，识别出在多实例资源系统中，循环是死锁的必要非充分条件。
8. Databases and Normalisation | 数据库与规范化
Entity-relationship diagrams and the translation to relational schemas are a rich source of marks. A notorious mistake is representing a many-to-many relationship without creating an associative entity (junction table). This leads to redundancy and anomalies in the resulting tables. Candidates also confuse cardinality and participation constraints, marking the wrong side of the relationship line.
实体关系图及其向关系模式的转换是一个得分富矿。一个众所周知的错误是表示多对多关系时未创建关联实体（连接表）。这会导致生成的数据表存在冗余和异常。考生还混淆基数约束和参与约束，把关系线的错误一侧标记出来。
The normalisation process up to Third Normal Form (3NF) is essential. A common error is removing partial dependencies without establishing a separate key for the new table, or leaving transitive dependencies unremoved because the non-key attribute is seen as 'important'. Every non-key attribute must depend on 'the key, the whole key, and nothing but the key' for 3NF. Providing a worked example that progresses from UNF to 1NF to 2NF to 3NF step by step is expected.
达到第三范式（3NF）的规范化过程至关重要。常见错误是消除部分依赖时未为新表建立单独的键，或因认为非键属性“重要”而保留传递依赖。对于3NF，每个非键属性必须依赖于“键、整个键且仅依赖于键”。应当提供从 UNF 到 1NF 到 2NF 再到 3NF 逐步进行的实例演示。
SQL questions test both data definition (DDL) and data manipulation (DML). An error is using DELETE instead of DROP TABLE to remove a table structure, or forgetting the WHERE clause in an UPDATE statement, which would modify all rows. In JOIN queries, Cartesian products arise when the join condition is omitted, leading to enormous result sets that are easy to spot but costly in exams.
SQL题目既测试数据定义（DDL）也测试数据操纵（DML）。一个错误是用DELETE代替DROP TABLE来删除表结构，或在UPDATE语句中忘记WHERE子句，从而修改所有行。在JOIN查询中，遗漏连接条件时会产生笛卡尔积，导致结果集庞大，虽容易发现但在考试中代价高昂。
9. Networking and Communication Protocols | 网络与通信协议
The TCP/IP stack and the encapsulation of data at each layer must be described accurately. A common misconception is stating that routers operate at the Data Link layer; routers operate at the Network layer and use IP addresses to forward packets, while switches use MAC addresses at the Data Link layer. The protocol data unit (PDU) names—segment, packet, frame—are interchanged carelessly.
TCP/IP协议栈及各层的数据封装必须准确描述。常见误解是称路由器工作在数据链路层；路由器工作在网络层并使用IP地址转发数据包，而交换机在数据链路层使用MAC地址。协议数据单元（PDU）的名称——段、包、帧——常被粗心互换。
ten label the sequence and acknowledge numbers incorrectly, or they forget to explain the significance of the initial sequence numbers being random to protect against session hijacking. Likewise, the four-step connection termination is less familiar, leading to incomplete answers about the TIME_WAIT state.
通过三次握手（SYN, SYN-ACK, ACK）建立TCP连接的过程是经典的图示题。考生常标错序列号和确认号，或忘记解释初始序列号随机分配的意义——为了防止会话劫持。类似地，四步连接终止过程较少被熟悉，导致关于TIME_WAIT状态的回答不完整。
Client-server and peer-to-peer models are compared for scalability, security, and fault tolerance. A pitfall is claiming that peer-to-peer networks are always more scalable without noting the trade-off in data consistency and the difficulty of searching without a centralised index. A balanced argument with real-world examples (e.g., BitTorrent vs. a web server) yields the highest marks.
客户端-服务器模型与对等网络模型在可扩展性、安全性和容错性方面进行比较。易错点是声称对等网络总是更具可扩展性，而未注意到数据一致性的权衡以及无集中索引时搜索的困难。用现实案例（如BitTorrent与Web服务器）进行平衡论证可获得最高分。
10. Computational Thinking and Problem Solving | 计算思维与问题求解
Algorithm design questions often present a scenario requiring abstraction, decomposition, and pattern recognition. The most common failure is jumping straight into pseudocode without first outlining the key steps or identifying edge cases. Samples with low marks show no pre-planning, resulting in logic that fails on boundary values such as empty lists or negative inputs.
算法设计题常设定一个需要抽象、分解和模式识别的情境。最常见的失败是直接跳到伪代码编写，而未先概述关键步骤或识别边界情况。低分样例没有预先规划，从而导致逻辑在空列表或负输入等边界值上失效。
Finite state machines (FSMs) and Turing machines are tested for both design and interpretation. When drawing an FSM, candidates forget to label the start state clearly or leave transitions undefined, which implies an implicit error state. According to the specification, all transitions must be defined, or the missing behaviour must be explained. A Mealy machine vs. Moore machine distinction should be explicit in the output timing.
有限状态机（FSM）和图灵机既考设计也考解释。在绘制FSM时，考生忘记清晰标注初始状态，或使得一些转换未定义，这意味着隐含的错误状态。按说明要求，所有转换必须定义，否则缺失的行为必须加以解释。在输出时序上，Mealy机和Moore机的区分应当明确。
Recursive problem solving is often attempted without a clear base case, causing infinite recursion or stack overflow. Even when a base case is present, it may not be reached due to an incorrectly defined recursive step. Writing a proof of termination by showing that the problem size reduces with each call is a strong examination technique.
递归求解问题常因没有明确的基例而尝试，导致无限递归或栈溢出。即使存在基例，也可能因递归步骤定义不正确而无法到达。通过展示每次调用时问题规模都减小来写出终止性证明，是一种强有力的考试技巧。
11. Moral, Ethical, Legal, and Environmental Issues | 道德、伦理、法律与环境问题
This section is often underestimated and answered superficially. To score highly, candidates must link general principles to specific legislation (e.g., GDPR, Computer Misuse Act, Copyright Designs and Patents Act) and technology contexts. A weak answer says 'privacy is important'; a strong answer explains how a relational database designer must implement data minimisation and purpose limitation to comply with GDPR Article 5.
这一部分常被低估且回答肤浅。要得高分，考生必须将一般原则与具体法规（如GDPR、计算机滥用法、版权设计和专利法）以及技术情境联系起来。较弱的答案是“隐私很重要”；优秀的答案则解释关系型数据库设计者必须如何实施数据最小化和目的限制以符合GDPR第5条。
When asked about the environmental impact of computing, common responses focus only on energy consumption and e-waste. Top-scoring scripts also discuss the positive effects: enabling smart grids, optimising logistics to reduce carbon footprint, and the dematerialisation of goods. The Pre-U mark scheme rewards nuanced evaluation.
被问及计算对环境的影响时，常见回答仅关注能耗和电子垃圾。高分答卷还会讨论积极效应：实现智能电网、优化物流以减少碳足迹，以及产品的非物质化。Pre-U评分方案奖励细致入微的评价。
12. Exam Technique and Common Blunders Across Papers | 跨题型考试技巧与常见失误
A recurring behaviour is not reading the command word carefully. 'Explain' requires a justification, not just a statement. 'Describe' requires a step-by-step account. Time management fails when candidates spend too long on an early low-mark question and then rush through a high-mark algorithm design. Allocating minutes proportionally to the available marks is non-negotiable.
一个反复出现的行为是没有仔细阅读指令词。“解释”需要给出理由，而不仅仅是一个陈述。“描述”需要逐步说明。当考生花过多时间在早期的低分题上，而匆匆完成高分算法设计题时，时间管理就失败了。按可用分值比例分配时间是无可妥协的。
Omitting units in calculation questions (e.g., kilobytes vs. kibibytes, milliseconds vs. microseconds) leads to lost precision marks. Similarly, leaving a final answer in an unsimplified form or with a floating point that should be normalised is penalised. In coding questions, not indenting pseudocode or missing out end-of-loop markers makes the logic unreadable and often incorrect.
在计算题中遗漏单位（例如kilobytes与kibibytes，milliseconds与microseconds）会导致丢失精度分。同样，将最终答案以未化简的形式或应该规格化的浮点数形式保留会被扣分。在编码题中，未缩进伪代码或遗漏循环结束标记会使逻辑难以阅读，且往往不正确。
Finally, very few candidates perform a final sanity check. For a calculated value, ask: does this magnitude make sense in context? For a logic diagram, does the truth table of the expression match the original? A five-minute review at the end can recover marks lost through simple slips like copying a number wrongly or circling the wrong K-map cell.
最后，极少数考生会进行最终的合理性检查。对于一个计算值，问一问：这个数量级在上下文中合理吗？对于一个逻辑图，这个表达式的真值表与原始表达式是否匹配？考试结束前花五分钟检查，就能挽回因抄错数字或圈错卡诺图单元格等简单失误而丢失的分数。
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