📚 Year 13 CIE Computer Science Core Concepts Revision | Year 13 CIE 计算机核心知识点梳理
Mastering Year 13 CIE Computer Science requires a solid grasp of both theoretical principles and their practical applications. This article revisits the essential topics that frequently appear across Paper 3 and Paper 4, from advanced data representation to object-oriented programming, databases, and security. Each section pairs concise English explanations with Chinese counterparts to reinforce bilingual understanding and aid revision for the final examination.
掌握 Year 13 CIE 计算机科学需要对理论原理及其实际应用有扎实的理解。本文重温 Paper 3 和 Paper 4 中频繁出现的核心主题，从高级数据表示到面向对象编程、数据库和安全。每个部分都以简明扼要的中英双语解释呈现，帮助巩固理解，为最终考试助力。
1. Advanced Data Representation | 高级数据表示
Floating-point numbers in binary are stored using a mantissa and an exponent, commonly in IEEE 754 format. The value is represented as ±1.m × 2^(e-127) for single precision, enabling a wide range but limited precision. Normalisation ensures a leading 1 bit after the binary point, maximising accuracy for a given bit pattern.
二进制浮点数使用尾数和指数存储，通常采用 IEEE 754 格式。单精度的值表示为 ±1.m × 2^(e-127)，能够表示很大范围但精度有限。规范化确保二进制小数点后有一个前导 1，从而在给定位数下最大化精度。
Bitmaps store images as a grid of pixels, each assigned a colour value using a certain bit depth. Higher bit depth yields more colours but increases file size. Vector graphics, by contrast, store geometric primitives and are resolution-independent, making them ideal for scalable illustrations.
位图将图像存储为像素网格，每个像素根据一定的位深度赋予颜色值。更高的位深度带来更丰富的色彩但增大文件尺寸。矢量图形则存储几何图元，与分辨率无关，非常适合可缩放插图。
Sound is captured through sampling: taking measurements of the wave’s amplitude at regular intervals. The sampling rate and bit depth determine audio quality and file size. Nyquist’s theorem states that the sampling rate must be at least twice the highest frequency to reconstruct the original signal accurately.
声音通过采样捕获：以固定间隔测量波形的幅度。采样率和位深度决定音质和文件大小。奈奎斯特定理指出，采样率必须至少是最高频率的两倍才能准确重建原始信号。
2. Communication and Internet Technologies | 通信与互联网技术
The TCP/IP protocol stack comprises four layers: Application (e.g., HTTP, FTP), Transport (TCP/UDP), Internet (IP), and Link. TCP provides reliable, connection-oriented delivery with sequencing and acknowledgements, while UDP offers low-latency, connectionless communication suitable for streaming.
TCP/IP 协议栈包含四层：应用层（如 HTTP、FTP）、传输层（TCP/UDP）、网络层（IP）和链路层。TCP 提供可靠、面向连接的传输，具有序号和确认机制；UDP 则提供低延迟、无连接通信，适用于流媒体。
Subnetting divides a network into smaller logical subnetworks using a subnet mask. The mask separates the network and host portions of an IP address. CIDR notation (e.g., 192.168.1.0/24) indicates the number of network bits, enabling efficient address allocation and routing.
子网划分利用子网掩码将网络划分为更小的逻辑子网。掩码分离出 IP 地址的网络部分和主机部分。CIDR 表示法（如 192.168.1.0/24）指明网络位的数量，实现高效的地址分配和路由。
Public IP addresses are globally unique and routable on the Internet, whereas private IP addresses are used within local networks and cannot be routed directly. Network Address Translation (NAT) allows multiple private IP devices to share one public IP for internet access, conserving IPv4 address space.
公共 IP 地址全球唯一，可在互联网上路由；私有 IP 地址则在本地网络内使用，不能直接路由。网络地址转换（NAT）允许多个私有 IP 设备共享一个公共 IP 上网，节省 IPv4 地址空间。
3. Processor Architecture and System Performance | 处理器架构与系统性能
Pipelining improves CPU throughput by overlapping instruction stages: fetch, decode, execute, and write-back. While one instruction is being executed, the next can be fetched. Hazards such as data dependency or branch misprediction can stall the pipeline, requiring techniques like forwarding and branch prediction.
流水线通过重叠指令的取指、译码、执行、写回阶段来提高 CPU 吞吐量。前一条指令正在执行时，下一条指令即可被取指。数据依赖或分支预测错误等冒险可能导致流水线停顿，需采用前推和分支预测等技术。
Multicore processors contain multiple independent cores on a single chip, enabling true parallel execution of threads or processes. Operating systems must support multitasking to utilise these cores efficiently. Software must be written with concurrency in mind to fully benefit from multicore architectures.
多核处理器在一个芯片上集成多个独立核心，真正实现线程或进程的并行执行。操作系统必须支持多任务处理以高效利用这些核心。软件编写时需考虑并发才能充分受益于多核架构。
Interrupts are signals from hardware or software that pause the current process so the CPU can handle urgent tasks. The processor saves the current state, executes an Interrupt Service Routine (ISR), then resumes. Interrupts enable responsive multitasking and real-time processing.
中断是来自硬件或软件的信号，用于暂停当前进程以便 CPU 处理紧急任务。处理器保存当前状态，执行中断服务程序（ISR）后恢复。中断实现了响应式多任务和实时处理。
4. System Software and Language Translators | 系统软件与语言翻译器
Operating systems manage hardware resources, provide a user interface, handle file management, and enforce security. Key functions include memory management (paging, segmentation), process scheduling, and input/output control. Virtual memory uses disk space to extend RAM, allowing larger programs to run.
操作系统管理硬件资源、提供用户界面、处理文件管理并实施安全策略。主要功能包括内存管理（分页、分段）、进程调度和输入输出控制。虚拟内存使用磁盘空间扩展 RAM，允许运行更大的程序。
Compilers translate high-level source code into machine code all at once, producing an executable file that runs faster. Interpreters translate and execute code line-by-line, better for debugging but slower in execution. Intermediate approaches like bytecode compilation (Java) combine portability with performance.
编译器一次性将高级源代码翻译成机器码，生成运行更快的可执行文件。解释器逐行翻译并执行，便于调试但执行较慢。字节码编译等中间方式（如 Java）兼顾了可移植性和性能。
The fetch-decode-execute cycle at the hardware level is orchestrated by the control unit, using registers such as PC, MAR, MDR, CIR, and ACC. Understanding how machine instructions are executed is fundamental to appreciating how assembly language and high-level code interact with the processor.
硬件层面的取指-译码-执行循环由控制单元协调，使用 PC、MAR、MDR、CIR 和 ACC 等寄存器。理解机器指令如何执行是理解汇编语言和高级代码与处理器交互的基础。
5. Security, Privacy and Data Integrity | 安全、隐私与数据完整性
Encryption converts plaintext into ciphertext using an algorithm and key. Symmetric encryption uses the same key for encryption and decryption, whereas asymmetric encryption employs a public-private key pair. Common applications include securing data at rest and in transit via protocols like TLS.
加密使用算法和密钥将明文转换为密文。对称加密使用相同密钥加解密，非对称加密则使用公私密钥对。常见应用包括通过 TLS 等协议保护静态数据和传输中的数据。
Digital signatures verify the authenticity and integrity of a message. The sender encrypts a hash of the message with their private key; the recipient decrypts it with the sender’s public key and compares the hash. If they match, the message is unaltered and genuinely from the sender.
数字签名验证消息的真实性和完整性。发送方使用自己的私钥加密消息的哈希值；接收方用发送方的公钥解密并比对哈希值。若匹配，则说明消息未被篡改且确实来自发送方。
SSL/TLS provides a secure channel between client and server through a handshake that establishes encryption parameters and authenticates the server (and optionally the client) using digital certificates. This protects sensitive data such as login credentials and payment information.
SSL/TLS 通过握手建立加密参数，并使用数字证书验证服务器（可选的客户端），从而在客户端和服务器之间建立安全通道。这保护了登录凭据和支付信息等敏感数据。
6. Ethics, Ownership and AI Implications | 伦理、所有权与人工智能影响
Ethical considerations in computing involve issues like data privacy, algorithmic bias, and the digital divide. AI systems can perpetuate existing biases if trained on skewed data. Developers have a responsibility to ensure fairness, transparency, and accountability in automated decision-making.
计算领域的伦理考量涉及数据隐私、算法偏见和数字鸿沟等问题。若在偏斜数据上训练，AI 系统可能延续既有偏见。开发人员有责任确保自动化决策的公平性、透明度和问责制。
Intellectual property rights protect software through copyright and licensing. Open-source licences (e.g., MIT, GPL) permit usage, modification, and distribution under defined terms, fostering collaborative development. Proprietary software restricts access to source code and often imposes usage limitations.
知识产权通过版权和许可保护软件。开源许可证（如 MIT、GPL）在明确条款下允许使用、修改和分发，促进协作开发。专有软件限制对源代码的访问，并常施加使用限制。
The impact of AI on employment, surveillance, and autonomous systems raises profound ethical questions. Legal frameworks like GDPR enforce data protection rights, while industry guidelines promote responsible AI deployment. Staying informed about these issues is part of a computer scientist’s professional duty.
人工智能对就业、监控和自主系统的影响引发深刻的伦理问题。GDPR 等法律框架强化数据保护权利，行业指南则促进 AI 的负责任部署。了解这些问题是一名计算机专业人员的职责所在。
7. Algorithm Design and Complexity | 算法设计与复杂度分析
Recursion solves problems by breaking them into smaller, self-similar subproblems and calling the same function until a base case is reached. Classic examples include factorial, Fibonacci, and tree traversals. Understanding the call stack and tail recursion optimisation helps prevent stack overflow.
递归通过将问题分解为更小的、自相似的子问题并调用同一函数，直到达到基线条件来解决问题。经典示例包括阶乘、斐波那契数列和树的遍历。理解调用栈和尾递归优化有助于防止栈溢出。
Time complexity is expressed using Big O notation, describing how the running time grows with input size n. For sorting algorithms, bubble sort is O(n²), merge sort is O(n log n), and binary search is O(log n). Space complexity is equally critical when memory is constrained.
时间复杂度使用大 O 表示法，描述运行时间随输入规模 n 增长的情况。排序算法中，冒泡排序为 O(n²)，归并排序为 O(n log n)，二分查找为 O(log n)。在内存受限时，空间复杂度同样关键。
Standard algorithm design paradigms include brute-force, divide-and-conquer, greedy, and dynamic programming. Dynamic programming stores intermediate results to avoid repeated calculations, transforming exponential problems into polynomial ones, as seen in the 0/1 knapsack and shortest path algorithms.
常见算法设计范式包括暴力法、分治法、贪心法和动态规划。动态规划存储中间结果以避免重复计算，将指数级问题转化为多项式级问题，如 0/1 背包和最短路径算法。
8. Data Structures and Their Implementations | 数据结构及其实现
Linked lists dynamically allocate nodes containing data and a pointer to the next node. Singly linked lists allow one-way traversal; doubly linked lists add a previous pointer. Insertion and deletion are O(1) at the head, but accessing an arbitrary element is O(n), unlike arrays with O(1) indexed access.
链表动态分配节点，内含数据及指向下一节点的指针。单链表允许单向遍历；双链表增加了前驱指针。在表头插入和删除为 O(1)，但访问任意元素为 O(n)，不像数组可 O(1) 索引访问。
Stacks follow Last-In-First-Out (LIFO) and support push/pop operations, while queues use First-In-First-Out (FIFO) with enqueue/dequeue. Priority queues dequeue the highest-priority element. These ADTs can be implemented using arrays or linked lists, each with trade-offs in efficiency.
栈遵循后进先出（LIFO），支持 push/pop 操作；队列采用先进先出（FIFO），具有入队和出队。优先队列出队优先级最高的元素。这些抽象数据类型可用数组或链表实现，各有优劣。
Binary trees consist of nodes with at most two children. Binary search trees (BST) maintain an ordered property: left child < parent < right child, enabling O(log n) search on average. Tree traversals – pre-order, in-order, post-order – visit nodes in different sequences, each serving specific applications.
二叉树由最多有两个子节点的节点构成。二叉搜索树（BST）保持有序性质：左子 < 父 < 右子，平均搜索时间为 O(log n)。树的遍历——前序、中序、后序——以不同顺序访问节点，各有特定应用。
9. Object-Oriented Programming (OOP) | 面向对象编程（OOP）
OOP organises code around objects that encapsulate data (attributes) and behaviour (methods). Classes define blueprints; instantiation creates objects. Encapsulation hides internal state through access modifiers (private, public), exposing only necessary interfaces, thus enhancing modularity and maintainability.
OOP 围绕封装了数据（属性）和行为（方法）的对象来组织代码。类定义蓝图；实例化创建对象。封装通过访问修饰符（私有、公有）隐藏内部状态，仅暴露必要接口，以增强模块性和可维护性。
Inheritance allows a subclass to reuse, extend, or override features of a superclass, promoting code reuse. Polymorphism lets objects of different classes be treated as instances of a common superclass, with method calls resolved at runtime via dynamic binding. Abstract classes and interfaces define contracts without full implementation.
继承允许子类复用、扩展或重写超类的特性，促进代码复用。多态使不同类的对象可被视为公共超类的实例，方法调用在运行时通过动态绑定解析。抽象类和接口定义契约，但不提供完整实现。
Design patterns such as singleton, factory, and observer capture best practices for recurring design problems. Applying them wisely leads to robust, decoupled code. In CIE contexts, students should be able to interpret UML class diagrams showing associations, aggregation, and inheritance.
单例、工厂和观察者等设计模式捕捉了常见设计问题的最佳实践。合理运用能带来稳健、解耦的代码。在 CIE 考试中，学生应能解读展示关联、聚合和继承的 UML 类图。
10. Database Systems and Normalisation | 数据库系统与规范化
Relational databases store data in tables linked by primary and foreign keys. SQL (Structured Query Language) is used to define, manipulate, and query data. Key SQL commands include SELECT, INSERT, UPDATE, DELETE, and JOIN operations that combine data from multiple tables.
关系数据库将数据存储在由主键和外键链接的表中。SQL（结构化查询语言）用于定义、操作和查询数据。主要的 SQL 命令包括 SELECT、INSERT、UPDATE、DELETE，以及用于合并多表数据的 JOIN 操作。
Normalisation reduces data redundancy and anomalies by decomposing tables into smaller, related tables. First Normal Form (1NF) removes repeating groups, 2NF removes partial dependencies on composite keys, and 3NF removes transitive dependencies. Denormalisation may be used for read-heavy systems to improve performance.
规范化通过将表分解为更小的相关表来减少数据冗余和异常。第一范式（1NF）移除重复组，第二范式（2NF）移除对组合键的部分依赖，第三范式（3NF）移除传递依赖。对于读密集型系统，可反规范化以提高性能。
Entity-Relationship (ER) diagrams model data entities, attributes, and relationships (one-to-one, one-to-many, many-to-many). They serve as a blueprint for database design, making it easier to identify tables, keys, and constraints before implementation.
实体-关系（ER）图对数据实体、属性和关系（一对一、一对多、多对多）进行建模，是数据库设计的蓝图，有助于在实施前确定表、键和约束。
11. Virtualisation and Cloud Computing | 虚拟化与云计算
Virtualisation abstracts physical hardware to create multiple virtual machines (VMs) running separate operating systems on one host. A hypervisor manages these VMs, enabling efficient resource utilisation, isolation, and easy migration. This underpins modern cloud infrastructure.
虚拟化将物理硬件抽象化，在一台主机上创建多个虚拟机（VM）运行独立的操作系统。虚拟机管理程序（Hypervisor）管理这些 VM，实现高效资源利用、隔离和便捷迁移，这是现代云基础设施的基础。
Cloud computing delivers on-demand IT resources over the internet: Infrastructure as a Service (IaaS) provides virtualised hardware; Platform as a Service (PaaS) adds runtime and middleware; Software as a Service (SaaS) delivers finished applications. Benefits include scalability, cost-effectiveness, and remote accessibility.
云计算通过互联网按需提供 IT 资源：基础设施即服务（IaaS）提供虚拟化硬件；平台即服务（PaaS）增加运行时和中间件；软件即服务（SaaS）交付最终应用程序。其优势包括可扩展性、成本效益和远程访问性。
Public, private, and hybrid cloud models offer different trade-offs in control, security, and cost. Understanding these distinctions is necessary for evaluating the architectural decisions made in large-scale system design.
公有云、私有云和混合云模式在控制权、安全性和成本方面各有权衡。理解这些差异对于评估大规模系统设计中的架构决策至关重要。
Published by TutorHao | Computer Science Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导