📚 High-Frequency Exam Topics & Common Mistakes Analysis | 高频考点与易错题分析
Pre-U CAIE Computer Science papers consistently test a core set of topics, yet many candidates lose marks on the same subtle details year after year. This article dissects the most frequently examined areas and the typical errors that can transform a grade A into a B. Understanding these patterns will sharpen your revision and help you avoid the pitfalls that trap even well-prepared students.
Pre-U CAIE 计算机科学试卷总是围绕一组核心主题出题，然而每年都有大量考生在同样的细节上丢分。本文深入剖析最高频的考点以及那些能将A等降为B等的典型错误。掌握这些规律后，你能更有针对性地复习，避开那些连准备充分的同学也常掉入的陷阱。
1. Data Representation & Binary Arithmetic | 数据表示与二进制运算
Questions on two's complement, fixed-point representation and floating-point normalisation appear almost every year. A classic blunder is forgetting that the range of an 8-bit two's complement integer is -128 to +127, not -127 to +127. When a candidate performs 80₁₆ + 80₁₆ (both negative in two's complement) and writes 00₁₆ without setting the overflow flag, they demonstrate a fundamental misunderstanding of wrap-around arithmetic.
补码、定点表示和浮点规格化几乎年年都考。一个经典错误是忘记 8 位补码整数的范围是 -128 到 +127，而非 -127 到 +127。当考生计算 80₁₆ + 80₁₆（两个数在补码下均为负值）后直接写下 00₁₆ 而不设置溢出标志，就暴露了对回绕算术的根本性误解。
Floating-point errors also surface when students try to represent 0.1 exactly in binary. They often waste time producing an infinite repeating pattern instead of recognising that the specification requires a standardised mantissa and exponent. In normalising, the mantra 'move the binary point until a single 1 appears before it' is simple, yet applying it to negative numbers in two's complement causes confusion: the mantissa must start with 10 for negative values, not 01.
浮点数错误同样常见，尤其是试图精确表示 0.1 的时候。很多同学花费大量时间写出无限循环的二进制模式，却未意识到考题要求的是标准化的尾数和阶码。规格化时，'移动二进制小数点直到前面仅有一个 1' 这条规则看似简单，但应用在补码负数上就会引起混乱：负数的尾数必须以 10 开头，而非 01。
2. Logic Gates & Boolean Algebra Simplification | 逻辑门与布尔代数化简
Karnaugh map grouping is a high-frequency skill that is easy to get wrong. A common blunder is grouping six cells as if they were a valid rectangle, or forgetting that the map wraps around both horizontally and vertically. When simplifying the expression for a seven-segment display decoder, students frequently miss the 'don't care' conditions, leading to circuits that are more complex than necessary and thus losing marks for inefficiency.
卡诺图的分组是一项高频技能，却极易出错。常见错误包括把六个单元格当作合法矩形来圈选，或者忘记卡诺图在水平和垂直方向都是首尾相连的。在简化七段数码管译码器的表达式时，学生常常忽略"无关项"条件，导致电路比需要的更复杂，最终因效率低下而丢分。
Another pitfall lies in converting a logic circuit back into a Boolean expression. Candidates may incorrectly write A.(B+C) when the gate arrangement clearly demands A.B + A.C. A single misreading of an OR gate's input bubble as an AND gate can cascade into an entirely wrong truth table. Always annotate intermediate outputs on the diagram before writing the final expression.
另一个陷阱是把逻辑电路转换回布尔表达式。考生有时会错误地写成 A.(B+C)，而门电路的排列分明要求的是 A.B + A.C。仅仅将或门上表示取反的小圆圈误读为与门，就可能层层传导出完全错误的真值表。务必先在电路图上标出各中间输出，再写出最终表达式。
3. Processor Architecture & the Fetch-Execute Cycle | 处理器架构与取指执行周期
The fetch-execute cycle is almost guaranteed to appear, often through a detailed description of register transfers. A typical mistake is stating that the Program Counter (PC) is incremented after the instruction is decoded. In reality, the PC is updated during the fetch stage so that it holds the address of the next instruction. Mixing this up loses crucial marks in state-trace questions.
取指执行周期几乎必考，通常要求详细描述寄存器间的传输。典型错误是说程序计数器 (PC) 在指令被译码后才递增。实际上，PC 在取指阶段就已经更新，以便存放下一条指令的地址。在状态追踪题中把顺序搞混会丢掉关键分数。
Interrupt handling is another high-yield area. Many students fail to state that the current contents of the PC must be saved to a stack or a special register before the Interrupt Service Routine (ISR) begins. The subsequent restore ensures program continuity. Not mentioning the push/pop mechanics or the disabling of further interrupts while the ISR executes reveals an incomplete mental model.
中断处理是另一个高频考点。许多学生没有点明，在执行中断服务程序 (ISR) 之前，必须先把 PC 的当前内容保存到栈或专用寄存器中，随后的恢复操作才能保证程序连续执行。若未提及压栈/弹栈机制，或未说明在执行 ISR 期间应屏蔽新的中断，则反映出心理模型的不完整。
4. Assembly Language & Addressing Modes | 汇编语言与寻址模式
Addressing modes (immediate, direct, indirect, indexed) combine both conceptual understanding and precise notation. An error as subtle as writing MOV R1, #5 instead of MOV R1, 5 for direct addressing costs marks because the # symbol denotes an immediate value. In indirect mode, a student might incorrectly claim the operand is the address of the value, whereas the register holds a pointer to the memory location containing the actual address.
寻址模式（立即、直接、间接、变址）既考概念理解，又考精确写法。一个细微的错误——比如在直接寻址时写成 MOV R1, #5 而不是 MOV R1, 5——就会丢分，因为 # 号表示立即数。在间接寻址中，有学生可能误称操作数就是数值的地址，实际上寄存器存放的是一个指针，指向存有实际地址的内存单元。
Another frequent slip is confusing the purpose of the CIR (Current Instruction Register) and the MDR (Memory Data Register) during a fetch operation. The instruction fetched from memory travels through the MDR before being placed in the CIR for decoding. Saying 'the instruction is directly loaded into the CIR from RAM' simplifies away a key step that examiners look for.
另一个常见疏漏是混淆取指操作中 CIR（当前指令寄存器）和 MDR（内存数据寄存器）的作用。从内存取出的指令先经过 MDR，再放入 CIR 供译码。如果说"指令直接从 RAM 加载到 CIR"，就省略了考官期望看到的关键步骤。
5. Data Structures: Stacks, Queues & Linked Lists | 数据结构：栈、队列与链表
k pointer. In a written answer, a candidate might present a pop procedure that decrements the pointer and returns the element, but fails to include an underflow test. Similarly, confusing a queue' ten leads to the wrong condition for fullness: (rear + 1) mod max = front, not rear = max.
栈和队列操作常以伪代码和纸面演算的形式考查。经典错误是未检查栈指针就从未空栈中弹出元素。在书面答案中，考生可能给出一个弹出过程——递减指针并返回元素，却没有包含下溢检测。类似地，混淆队列的线性实现和循环实现，往往会导致错误地写满条件：(rear + 1) mod max = front，而非 rear = max。
Linked list manipulation problems demand meticulous pointer updates. A frequent error when inserting a node between A and B is to make A.next point to the new node before setting new.next to B. This destroys the only link to B and orphans the tail of the list. Drawing box-and-pointer diagrams is essential to avoid such sequencing errors.
链表操作题要求极其严谨的指针修改。在 A 和 B 之间插入结点时，一个常见错误是先将 A.next 指向新结点，再设置新结点的 next 为 B。这会破坏指向 B 的唯一链接，导致链表尾部被孤立。绘制框-指针图对于避免此类顺序错误至关重要。
6. Trees & Graph Traversal Algorithms | 树与图的遍历算法
Pre-order, in-order and post-order traversals of binary trees are a staple. The standard blunder is to output the root too early. For a tree representing the expression (A+B)*C, the correct post-order is A B + C * . Students who produce A B C + * have performed a misguided hybrid traversal. Similarly, applying pre-order to a binary search tree and claiming the output is sorted reveals confusion between BST properties and traversal order.
二叉树的前序、中序和后序遍历是必考点。典型错误是过早输出根结点。对于表示表达式 (A+B)*C 的树，正确的后序是 A B + C *。而写出 A B C + * 的同学实际上是进行了一种不正确的混合遍历。同样，对二叉搜索树进行前序遍历却声称输出是有序的，这暴露出对 BST 性质和遍历顺序的混淆。
Graph algorithms such as Dijkstra's shortest path and the A* heuristic bring further pitfalls. A common one is updating the distance of a vertex without re-checking its neighbours. Candidates sometimes fail to record the path itself, only outputting the final distance. Examiner mark schemes explicitly require the full sequence of steps or the final route, not just the numeric cost.
图算法如 Dijkstra 最短路和 A* 启发式同样暗藏陷阱。常见的错误是更新某个顶点的距离后却没有重新检查其邻居。考生有时只输出最终距离，忘记了记录路径本身。阅卷评分标准明确要求给出完整步骤序列或最终路线，而不仅仅是数字代价。
7. Algorithm Design & Big-O Notation | 算法设计与大O记法
Complexity analysis questions demand that students distinguish between best, average and worst cases. A recurring error is to label a linear search's worst case as O(n/2). The notation drops constants, so the correct answer is O(n). Similarly, claiming that binary search on a sorted array is O(log₂ n) in the worst case is correct, but writing O(log n) without specifying base is acceptable; however, saying 'always faster than linear search' ignores constant factors for small n.
复杂度分析题要求学生区分最好、平均和最坏情况。一个反复出现的错误是把线性查找的最坏情况标为 O(n/2)。大O记法要去掉常数，正确答案是 O(n)。类似地，声称对有序数组的二分查找最坏情况为 O(log₂ n) 是正确的，写 O(log n) 而不指定底数也可以接受；但如果断言“它总是比线性查找快”，则忽略了小规模 n 下的常数因子。
that calls itself with n-1 but lacks a check if n=0 results in infinite recursion. The exam often asks to trace such erroneous code, and students must articulate why the call stack eventually overflows. Visualising the call stack growth is an essential debugging skill.
递归算法设计是另一个高频领域，基例要么遗漏，要么放错位置。一个递归阶乘函数调用自身时使用 n-1，却没检查 n=0 的情况，就会导致无限递归。考试常要求追踪此类错误代码，学生必须解释为何调用栈最终会溢出。可视化调用栈的增长是一项关键调试技能。
8. Programming Paradigms & Parameter Passing | 编程范型与参数传递
Pre-U papers often include short Pascal-like pseudocode and examine the distinction between call-by-value and call-by-reference. A typical error is to assume that changes to a reference parameter inside a procedure have no effect on the argument in the calling environment. Conversely, when a large data structure must not be modified, passing it by value inadvertently creates a complete copy, wasting memory and demonstrating a lack of understanding of efficiency vs. safety trade-offs.
Pre-U 试卷经常给出简短的类 Pascal 伪代码，考查值传递与引用传递的区别。典型错误是以为在过程内部修改引用参数不会影响调用环境中的实参。反过来，当某个大型数据结构本不应被修改时，却使用了值传递，无意间创建了一份完整副本，既浪费内存，又暴露出对效率与安全性取舍的理解不足。
Scope rules also trap the unwary. A local variable in a function hides a global variable of the same name. Students frequently write 'the global variable is destroyed' or 'it becomes inaccessible permanently'. The correct description is that the local variable has precedence within its block; the global variable retains its value and becomes accessible again once the function returns, provided no explicit modification occurred.
作用域规则同样会让粗心的考生中招。函数内的局部变量会隐藏同名的全局变量。学生常写"全局变量被销毁"或"永久不可访问"。正确的描述是：在其块内，局部变量优先；全局变量仍然保留其值，在函数返回后便可重新访问，前提是没有对它进行过显式修改。
9. Databases: Normalisation & SQL | 数据库：规范化与SQL
Normalisation to Third Normal Form (3NF) is a topic where partial and transitive dependencies are frequently misidentified. A common mistake is to leave a non-key attribute that depends on part of a composite key in the same table, claiming it is in 2NF. Equally, students often split tables unnecessarily, creating over-normalised schemas that impair performance without eliminating any actual dependency violation.
将数据库规范到第三范式 (3NF) 是容易误判部分依赖和传递依赖的主题。常见错误是把一个依赖于复合主键一部分的非键属性留在同一张表中，还声称其满足 2NF。同样，学生经常不必要地分割表格，形成过度规范化的模式，损害性能却并未消除任何真正的依赖违规。
SQL queries, especially those involving GROUP BY and aggregate functions, account for significant marks. A classic pitfall is to write SELECT name, MAX(salary) FROM employees; without a GROUP BY clause, this usually triggers an error in most DBMS, yet candidates assume it returns the employee with the highest salary. The correct approach uses a subquery or a HAVING clause. Joins also cause trouble: confusing INNER JOIN with LEFT JOIN leads to missing rows that should appear when there is no matching record.
SQL 查询，尤其是涉及 GROUP BY 和聚合函数的查询，占分很重。一个经典陷阱是写出 SELECT name, MAX(salary) FROM employees; 而没有 GROUP BY 子句，这在多数 DBMS 中都会报错，考生却以为这样会返回工资最高的员工。正确的做法是使用子查询或 HAVING 子句。连接操作也容易出问题：混淆 INNER JOIN 和 LEFT JOIN 会导致在无匹配记录时本应出现的行被遗漏。
10. Networking & Cybersecurity Essentials | 网络与网络安全基础
The TCP/IP stack and the role of protocols at each layer are heavily tested. Many students misunderstand the difference between TCP and UDP, stating 'TCP is faster' without qualification. When asked to explain packet switching, a common error is to describe circuit switching instead, detailing a dedicated path being set up before transmission. The correct model involves packets taking independent routes and being reassembled at the destination.
TCP/IP 协议栈以及各层协议的作用考查频繁。许多学生误解了 TCP 与 UDP 的区别，不加限定地说"TCP 更快"。当要求解释分组交换时，常见错误却是描述电路交换，详细说明传输前建立了一条专用路径。正确的模型则是数据包沿着独立路径传输，在目的地重新组装。
In cybersecurity, symmetric vs. asymmetric encryption is a favourite comparison. A subtle blunder is to claim that asymmetric encryption is always more secure. In practice, both can be equally secure given sufficiently long keys; the real distinction lies in key distribution and computational cost. Similarly, digital signatures are often confused with simple encryption: a digital signature involves encrypting a message digest with the sender's private key, not encrypting the entire message.
在网络安全方面，对称与非对称加密的比较是常考内容。一个微妙错误是断言非对称加密总是更安全。实际上，只要密钥足够长，两者都可以同样安全；真正的区别在于密钥分发和计算开销。类似地，数字签名常与普通加密混淆：数字签名是用发送者的私钥加密消息摘要，而不是加密整个消息。
11. Object-Oriented Programming & UML | 面向对象编程与UML
Pre-U questions now regularly feature class diagrams and the implementation of inheritance. A high-frequency error is to write a constructor for a subclass that does not explicitly call the superclass constructor, yet the mark scheme demands a 'super()' or equivalent call. In theoretical discussions, candidates frequently misuse the 'is-a' and 'has-a' relationships, modelling a car as a subclass of engine (should be composition) instead of vehicle (inheritance).
Pre-U 试题现在经常会涉及类图和继承的实现。一个高频错误是编写子类构造函数时未显式调用超类构造函数，但评分标准却要求有 'super()' 或等同的调用。在理论讨论中，考生常误用"is-a"和"has-a"关系，将汽车建模为引擎的子类（应该用组合），而非交通工具的子类（继承）。
Polymorphism and method overriding also trip up candidates. When a parent reference points to a child object, the executed method depends on the actual object type (dynamic binding), not the reference type. Students incorrectly predict that the parent's version will run, losing marks on tracing questions. Ensure you understand which language feature (virtual methods in Pascal/C++ or default in Java) determines this behaviour.
多态和方法重写也常让考生栽跟头。当父类引用指向子类对象时，执行的方法取决于实际对象类型（动态绑定），而不是引用类型。学生错误地预测父类的版本会运行，从而在追踪题中丢分。务必弄明白到底是哪个语言特性（Pascal/C++ 中的 virtual 方法或 Java 中的默认行为）决定了这一行为。
12. Software Engineering & the Development Lifecycle | 软件工程与开发生命周期
The waterfall model, Agile, and Rapid Application Development (RAD) are compared on almost every paper. A typical mistake is to assert that agile methods perform no documentation at all. The mark scheme expects a balanced view: documentation is 'lightweight' or 'just-in-time', not absent. Another weak answer is stating that the waterfall model is 'bad' because changes are difficult, without acknowledging situations where requirements are stable and well-understood, making waterfall perfectly suitable.
瀑布模型、敏捷和快速应用开发 (RAD) 几乎每张试卷都会进行比较。典型错误是声称敏捷方法完全不产生任何文档。评分标准期望的是平衡的观点：文档是"轻量级"或"即时"的，而非完全缺失。另一个薄弱答案是声称瀑布模型"不好"，因为难以应对变化，却忽略了在需求稳定且充分理解的场景下，瀑布模型其实非常合适。
Testing strategies too are a source of common error. Many candidates describe alpha testing as 'done by the development team internally' and beta testing as 'done by the public', which is broadly correct, but they forget to mention that alpha testing typically uses white-box techniques while beta testing is black-box. Additionally, when writing test plans, students often neglect to specify expected outcomes for each test case, leaving only the input data.
测试策略同样是常见错误来源。许多考生描述 α 测试是"由开发团队内部完成"，β 测试是"由公众完成"，这大致正确，但他们忘了提及 α 测试通常采用白盒技术，而 β 测试是黑盒的。此外，在编写测试计划时，学生常常忽略为每个测试用例指定预期结果，只留下输入数据。
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