📚 Essential A-Level Computer Science Vocabulary | A-Level 计算机科学核心术语速记指南
Mastering technical vocabulary is fundamental to excelling in Cambridge Year 13 Computer Science (A2). This guide presents high-frequency terms organised by topic, with clear explanations and memory shortcuts that connect theory to real computing practice. Use it as a revision checklist and a quick-reference companion when tackling past papers or describing algorithms.
掌握专业术语是在剑桥 Year 13（A2）计算机科学中取得优异成绩的基础。本指南按主题整理高频术语，提供清晰易懂的说明与记忆捷径，将理论与实际计算相结合。你可以把它当作复习清单，也可以在做真题或描述算法时作为速查伴侣。
1. Data Representation and Structures | 数据表示与数据结构
Abstract Data Type (ADT) – a logical description of a data structure that specifies the operations possible without revealing implementation details. Example: a stack ADT defines push and pop, but the underlying storage could be an array or linked list.
抽象数据类型 (ADT) – 数据结构的逻辑描述，指明可以执行哪些操作，但不暴露实现细节。例如，栈 ADT 定义了 push 和 pop，但底层存储可以是数组也可以是链表。
Dynamic vs. Static Data Structure – a dynamic structure can grow or shrink at runtime (e.g., linked list, binary tree), while a static structure has a fixed size determined at compile time (e.g., array).
动态与静态数据结构 – 动态结构在运行时可以增长或收缩（如链表、二叉树），静态结构的大小在编译时就已固定（如数组）。
Stack Frame – a contiguous block of memory containing return addresses, parameters, and local variables, created each time a subroutine is called. It is fundamental to understanding recursion and exception handling.
栈帧 – 一块连续的内存区域，包含返回地址、参数和局部变量，每次调用子程序时创建。理解栈帧是理解递归和异常处理的基础。
Hash Table Collision Resolution – methods to handle two keys mapping to the same index: chaining (storing a linked list at each slot) or open addressing (probing for the next free slot).
哈希表冲突解决 – 处理两个键映射到同一索引的方法：链地址法（每个槽位存储链表）或开放寻址法（探测下一个空槽）。
B-tree and B+ tree – self-balancing tree structures optimised for disk storage, where nodes can have more than two children. B+ trees store all data in leaf nodes linked sequentially, making range queries efficient.
B 树与 B+ 树 – 为磁盘存储优化的自平衡树结构，节点可拥有两个以上子节点。B+ 树将所有数据存储在顺序链接的叶节点中，使范围查询更高效。
2. Algorithms and Complexity | 算法与复杂度
Big-O Notation (O) – describes the upper bound of an algorithm’s time or space complexity as input size n grows. Common classes: O(1), O(log n), O(n), O(n log n), O(n²), O(2ⁿ).
大 O 标记法 (O) – 描述随着输入规模 n 增长，算法时间复杂度或空间复杂度的上界。常见类别：O(1)、O(log n)、O(n)、O(n log n)、O(n²)、O(2ⁿ)。
Divide and Conquer – algorithmic paradigm that recursively breaks a problem into smaller subproblems, solves them independently, and combines the results (e.g., merge sort, quicksort).
分治法 – 将问题递归分解为更小的子问题、独立求解再合并结果的算法范式（例如归并排序、快速排序）。
Backtracking – a refinement of brute-force search that abandons partial candidates as soon as it determines they cannot lead to a valid solution. Used in puzzles, maze solving, and constraint satisfaction (e.g., depth-first branch and bound).
回溯法 – 对蛮力搜索的优化，一旦确定当前部分候选解不可能产生有效解，便立即放弃。用于谜题、迷宫求解和约束满足问题（如深度优先分支定界）。
Heuristic – a problem-solving technique that uses a “rule of thumb” to produce a good enough solution quickly when an exact solution is impractical. In A* search, the heuristic function estimates the cost from a node to the goal.
启发式 – 当精确解不可行时，利用“经验法则”快速得到足够好解的方法。在 A* 搜索中，启发函数用于估算从节点到目标的代价。
Halting Problem – the undecidable problem of determining, from a description of an arbitrary program and an input, whether the program will finish running or continue to loop forever. Proves that some problems cannot be solved by any algorithm.
停机问题 – 不可判定问题：给定任意程序及其输入，判断程序是否会终止运行。证明了有些问题任何算法都无法解决。
3. Programming Paradigms and Translation | 编程范式与翻译
Imperative vs. Declarative – imperative programming focuses on how to do something through sequences of commands (C, Java). Declarative programming specifies what the result should be (SQL, HTML, functional languages).
命令式与声明式 – 命令式编程侧重通过一系列命令来描述如何做（C、Java）。声明式编程指定结果应是什么（SQL、HTML、函数式语言）。
Low-level vs. High-level Language – low-level languages (assembly, machine code) are closely tied to hardware; high-level languages (Python, Java) use abstractions closer to human thought. Translators bridge the gap: assembler, compiler, interpreter.
低级语言与高级语言 – 低级语言（汇编、机器码）与硬件紧密相关；高级语言（Python、Java）使用接近人类思维的抽象。翻译程序弥合差距：汇编器、编译器、解释器。
Compiler, Interpreter, and JIT – a compiler translates the entire source code into machine code before execution; an interpreter executes line by line; Just-In-Time compilation (JIT) combines both by compiling frequently executed bytecode sections at runtime.
编译器、解释器与即时编译 – 编译器在执行前将全部源代码翻译为机器码；解释器逐行执行；即时编译 (JIT) 结合两者，在运行时编译频繁执行的字节码段。
Lexical Analysis, Syntax Analysis, Code Generation – the typical phases of a compiler: tokenisation (lexing), parsing into an abstract syntax tree, and generating target code. Semantic analysis checks meaning (type checking).
词法分析、语法分析、代码生成 – 典型编译阶段：标记化（词法分析）、解析为抽象语法树、生成目标代码。语义分析检查含义（类型检查）。
Backus-Naur Form (BNF) – a notation for context-free grammars used to specify the syntax of programming languages. Example: <digit> ::= 0|1|2|3|4|5|6|7|8|9.
巴科斯-诺尔范式 (BNF) – 用于描述上下文无关文法的符号系统，用来规定编程语言的语法。例如：<数字> ::= 0|1|2|3|4|5|6|7|8|9。
4. Computer Architecture and Hardware | 计算机体系结构与硬件
Von Neumann Architecture vs. Harvard Architecture – Von Neumann stores instructions and data in a shared memory, leading to the “Von Neumann bottleneck.” Harvard uses separate memories, allowing parallel access – common in RISC microcontrollers and DSPs.
冯·诺依曼架构 vs. 哈佛架构 – 冯·诺依曼将指令和数据存放在共享存储器中，导致“冯·诺依曼瓶颈”。哈佛架构用分离的存储器，允许并行访问——常用于 RISC 微控制器和数字信号处理器。
Pipelining – a processor technique where multiple instructions are overlapped in execution (fetch, decode, execute, write-back). A flush is needed on branch mispredictions.
流水线 – 处理器技术，多条指令在取指、译码、执行、写回阶段重叠执行。分支预测错误时需要进行流水线冲刷。
RISC vs. CISC – Reduced Instruction Set Computing uses a small, simple set of instructions emphasising software; Complex Instruction Set Computing has many powerful, multi-step instructions aiming to reduce code size. Modern processors often blend both.
精简指令集 vs. 复杂指令集 – RISC 采用少量简单指令，侧重软件；CISC 拥有众多强大的多步指令，旨在减少代码量。现代处理器常融合两者。
Cache Memory (L1, L2, L3) – high-speed memory located close to the CPU that stores frequently accessed data. Cache hit/miss rates and replacement policies (LRU, FIFO) are critical “Cambridge topics.”
高速缓存 (L1、L2、L3) – 靠近 CPU 的高速存储器，存放频繁访问的数据。缓存命中/缺失率和替换策略（LRU、FIFO）是剑桥常考主题。
Interrupt and ISR – an interrupt is a signal to the processor to suspend current execution and jump to an Interrupt Service Routine. Vectored interrupts provide the ISR address directly, prioritisation resolves simultaneous requests.
中断与中断服务例程 (ISR) – 中断是发送给处理器的信号，要求暂停当前执行并跳转到中断服务例程。向量中断直接提供 ISR 地址，优先级用来解决同时请求。
5. Operating Systems and Resource Management | 操作系统与资源管理
Multitasking: Preemptive vs. Cooperative – preemptive multitasking (modern OS) forces tasks to yield the CPU after a time slice; cooperative relies on tasks voluntarily yielding. Preemptive avoids deadlock from greedy processes.
多任务：抢占式与协作式 – 抢占式多任务（现代操作系统）在时间片后强制任务让出 CPU；协作式依赖任务主动让出。抢占式可避免因贪婪进程而导致的死锁。
Virtual Memory and Paging – virtual memory allows execution of processes not completely in physical RAM by mapping virtual addresses to physical frames. Pages are swapped using demand paging and page replacement algorithms.
虚拟内存与分页 – 虚拟内存通过将虚拟地址映射到物理帧，使得没有完全驻留在物理 RAM 中的进程也能执行。页面通过请求分页和页面置换算法进行交换。
Deadlock Conditions – mutual exclusion, hold and wait, no preemption, and circular wait. Strategies: prevention, avoidance (Banker’s algorithm), detection, and recovery.
死锁条件 – 互斥、持有并等待、不可抢占、循环等待。处理策略：预防、避免（银行家算法）、检测与恢复。
Semaphore and Mutex – a semaphore is an integer variable accessed through wait (P) and signal (V) atomic operations used for process synchronisation. A mutex (mutual exclusion lock) is a binary semaphore guarding a critical section.
信号量与互斥锁 – 信号量是一个整型变量，通过 P (wait) 和 V (signal) 原子操作实现进程同步。互斥锁是保护临界区的二元信号量。
6. Data Communication and Networking | 数据通信与网络
TCP/IP Stack Layers – Application (HTTP, FTP, SMTP), Transport (TCP, UDP), Internet (IP, routing protocols), Link (Ethernet, Wi-Fi). Know the encapsulation and decapsulation process.
TCP/IP 协议栈分层 – 应用层 (HTTP、FTP、SMTP)、传输层 (TCP、UDP)、网络层 (IP、路由协议)、链路层 (以太网、Wi-Fi)。需掌握封装与解封过程。
Packet Switching vs. Circuit Switching – packet switching divides data into packets that travel independently through a shared network; circuit switching establishes a dedicated physical path (e.g., traditional telephone network).
分组交换与电路交换 – 分组交换将数据分割为独立传输的数据包，通过共享网络传送；电路交换建立专用物理通路（如传统电话网）。
Subnet Mask and CIDR – a subnet mask determines the network prefix of an IP address. Classless Inter-Domain Routing (CIDR) notation: 192.168.1.0/24 means 24-bit network prefix.
子网掩码与无类域间路由 (CIDR) – 子网掩码确定 IP 地址的网络前缀。CIDR 表示法：192.168.1.0/24 表示 24 位网络前缀。
DNS Resolution – the hierarchical process that translates a domain name to an IP address, involving recursive resolvers, root servers, TLD servers, and authoritative name servers. Caching reduces latency.
DNS 解析 – 将域名转换为 IP 地址的层次化过程，涉及递归解析器、根服务器、顶级域服务器和权威名称服务器。缓存可降低延迟。
Symmetric vs. Asymmetric Encryption – symmetric uses the same key for encryption and decryption (AES, DES); asymmetric uses a public/private key pair (RSA). Hybrid approaches use asymmetric to exchange a symmetric session key.
对称加密与非对称加密 – 对称加密使用相同密钥加解密 (AES、DES)；非对称加密使用公/私钥对 (RSA)。混合方法用非对称加密交换对称会话密钥。
7. Databases and SQL | 数据库与 SQL
Normalisation (1NF, 2NF, 3NF) – a process to reduce data redundancy by organising attributes into relations: 1NF eliminates repeating groups; 2NF removes partial dependencies on a composite key; 3NF removes transitive dependencies.
规范化 (1NF、2NF、3NF) – 通过将属性组织成关系来减少数据冗余的过程：1NF 消除重复组；2NF 消除对组合键的部分依赖；3NF 消除传递依赖。
ACID Properties – Atomicity, Consistency, Isolation, Durability. These guarantee that database transactions are processed reliably even in the event of system failure.
ACID 特性 – 原子性、一致性、隔离性、持久性。这些特性保证即使在系统故障时，数据库事务也能可靠处理。
Entity-Relationship Diagram (ERD) – a visual representation of database entities, their attributes, and relationships (one-to-one, one-to-many, many-to-many). Foreign keys implement relationships in relational schemas.
实体关系图 (ERD) – 数据库实体、属性及其关系（一对一、一对多、多对多）的可视化表示。外键在关系模式中实现联系。
Stored Procedure and Trigger – a stored procedure is a precompiled collection of SQL statements stored on the server. A trigger is a special type of stored procedure that automatically executes in response to certain events (INSERT, UPDATE, DELETE).
存储过程与触发器 – 存储过程是编译后保存在服务器上的一组 SQL 语句。触发器是一种特殊的存储过程，当发生特定事件 (INSERT, UPDATE, DELETE) 时自动执行。
8. System Software and Security | 系统软件与安全
Malware Types – virus (attaches to clean files), worm (self-replicates across network), Trojan (masquerades as legitimate software), ransomware (encrypts data for ransom), spyware (stealthily collects information).
恶意软件类型 – 病毒（附着在干净文件上）、蠕虫（通过网络自我复制）、木马（伪装成合法软件）、勒索软件（加密数据索要赎金）、间谍软件（暗中收集信息）。
Firewall and Proxy Server – a firewall filters incoming and outgoing network traffic based on security rules. A proxy server acts as an intermediary, hiding internal IP addresses and often caching web content.
防火墙与代理服务器 – 防火墙根据安全规则过滤进出网络流量。代理服务器充当中介，隐藏内部 IP 地址，常会缓存网页内容。
Digital Signature and Certificate – a digital signature is created by encrypting a hash of a message with the sender’s private key, verifying integrity and authenticity. A digital certificate binds a public key to an identity, issued by a Certificate Authority (CA).
数字签名与数字证书 – 数字签名用发送者私钥加密消息哈希值，验证完整性和真实性。数字证书将公钥与身份绑定，由证书颁发机构 (CA) 签发。
Buffer Overflow and Injection Attacks – buffer overflow occurs when a program writes more data to a buffer than it can hold, potentially overwriting adjacent memory. SQL injection inserts malicious SQL via input fields. Both require input validation.
缓冲区溢出与注入攻击 – 缓冲区溢出指程序向缓冲区写入超出其容量的数据，可能覆盖相邻内存。SQL 注入通过输入字段插入恶意 SQL。两者均需输入验证。
9. Boolean Algebra and Logic Circuits | 布尔代数与逻辑电路
Minterm and Maxterm – a minterm is a product (AND) term where each variable appears exactly once in true or complemented form; a maxterm is a sum (OR) term. They help construct sum-of-products and product-of-sums expressions.
最小项与最大项 – 最小项是每个变量都恰好以原变量或反变量形式出现一次的乘积 (AND) 项；最大项是和 (OR) 项。它们用于构造与-或式和或-与式。
Karnaugh Map (K-map) – a graphical method to simplify Boolean expressions by grouping adjacent cells containing 1s. Groups of 2, 4, 8 eliminate variables and yield minimal logic.
卡诺图 (K-map) – 通过将含 1 的相邻单元格分组来简化布尔表达式的图形方法。2、4、8 个一组的组合能消去变量，得到最简逻辑。
Flip-flop Types – SR (Set-Reset), JK (toggles when J=K=1, with clock), D (data latch), T (toggle). Level-triggered vs. edge-triggered behaviour is essential for sequential logic design.
触发器类型 – SR（置位-复位）、JK（J=K=1 时翻转，有时钟）、D（数据锁存）、T（翻转）。电平触发与边沿触发行为对时序逻辑设计至关重要。
Half-adder and Full-adder – a half-adder adds two bits producing Sum and Carry. A full-adder adds three bits (including carry-in). Combine full-adders to build a ripple carry adder.
半加器与全加器 – 半加器将两个比特相加，产生和与进位；全加器将三个比特相加（包含进位输入）。级联全加器可构成行波进位加法器。
10. Internet Technologies and Protocols | 互联网技术与协议
HTTP Request Methods – GET (retrieve resource), POST (submit data), HEAD (headers only), PUT (update/replace), DELETE (remove). Stateless protocol: each request is independent; cookies/sessions maintain state.
HTTP 请求方法 – GET（获取资源）、POST（提交数据）、HEAD（仅首部）、PUT（更新/替换）、DELETE（删除）。无状态协议：每次请求独立；Cookie/会话维持状态。
REST and CRUD – Representational State Transfer (REST) uses standard HTTP methods to perform Create, Read, Update, Delete operations on resources identified by URLs. JSON is the common data interchange format.
REST 与 CRUD – 表述性状态传递 (REST) 使用标准 HTTP 方法对以 URL 标识的资源执行创建、读取、更新、删除操作。JSON 是常用数据交换格式。
Packet TTL and Fragmentation – Time-To-Live is decremented by each router; when it reaches 0 the packet is dropped. Fragmentation occurs when a packet exceeds the Maximum Transmission Unit (MTU) of a link; fragments are reassembled at the destination.
数据包 TTL 与分片 – 生存时间每经过一个路由器递减；归零则丢弃。当数据包超过链路最大传输单元 (MTU) 时发生分片，碎片在目的地重组。
DHCP and NAT – Dynamic Host Configuration Protocol assigns IP addresses and network parameters automatically. Network Address Translation maps multiple private IP addresses to a single public IP, conserving IPv4 space.
DHCP 与 NAT – 动态主机配置协议自动分配 IP 地址和网络参数。网络地址转换将多个私有 IP 映射到单个公网 IP，节省 IPv4 地址。
11. Artificial Intelligence and Machine Learning Basics | 人工智能与机器学习基础
Supervised, Unsupervised, Reinforcement Learning – supervised uses labelled data to predict outcomes (regression, classification); unsupervised finds hidden patterns in unlabelled data (clustering); reinforcement learns by trial and error interaction with an environment to maximise a reward signal.
监督学习、无监督学习与强化学习 – 监督学习用标注数据预测结果（回归、分类）；无监督学习在未标注数据中发现隐藏模式（聚类）；强化学习通过与环境交互试错，最大化奖励信号。
Neural Network Backpropagation – the algorithm that computes gradients of the loss function with respect to weights, propagating errors backward through the network, enabling gradient descent optimisation.
神经网络反向传播 – 计算损失函数对权重的梯度，将误差沿网络反向传播，从而实现梯度下降优化的算法。
Overfitting and Regularisation – overfitting occurs when a model learns noise in training data, failing to generalise. Regularisation techniques (dropout, L1/L2 penalty) constrain complexity to improve validation performance.
过拟合与正则化 – 过拟合指模型学习了训练数据中的噪声，无法泛化。正则化技术（Dropout、L1/L2 惩罚）限制模型复杂度，提升验证性能。
Published by TutorHao | Cambridge 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