OCR GCSE Computer Science 算法 编程 备考技巧
OCR GCSE Computer Science (J277) 是英国 OCR 考试局的计算机科学 GCSE 课程，两份笔试各占 50%，满分都是 80 分，考试时长各 1 小时 30 分钟。很多同学觉得这门课"背一背就能过"，结果在大题上大面积丢分：算法题被扣过程分，编程题写不出伪代码，网络安全题只会写一个单词。这篇文章用考点拆解的方式，把算法、编程、数据表示、计算机系统与网络安全五大板块的核心知识和提分技巧一次讲清。
OCR GCSE Computer Science (J277) is assessed by two written papers, each worth 50% of the qualification and each worth 80 marks with a duration of 1 hour 30 minutes. Many students assume the subject can be passed by memorising definitions, then lose large numbers of marks on extended questions: they miss method marks on algorithm traces, they cannot write correct pseudocode, and they answer network security questions with a single keyword. This guide breaks the specification into the five areas that actually decide your grade: algorithms, programming, data representation, computer systems, and network security.
一、先看清考纲：Paper 1 与 Paper 2 的分工
Paper 1 是 Computer systems，考 CPU、内存、存储、网络、系统软件、伦理与法律问题，几乎全是"解释、说明、比较"类题型。Paper 2 是 Computational thinking, algorithms and programming，考算法思维、伪代码、搜索排序、编程技巧、布尔逻辑、语言与 IDE，还包含一道 8 分的编程大题。两份卷子的共同点是：定义题只占少数，绝大多数分值落在"为什么"和"怎么做"上。因此，光背术语拿不到 7 分以上，必须练解释性表达。
Paper 1, Computer systems, covers the CPU, memory, storage, networks, system software, and ethical and legal issues. Almost every question is an explain, describe, or compare question. Paper 2, Computational thinking, algorithms and programming, covers computational thinking, pseudocode, searching and sorting, programming techniques, Boolean logic, languages and IDEs, and includes an extended programming question worth 8 marks. In both papers, definition questions are a minority of the marks, and the majority of the marks sit in the why and how questions. Memorising vocabulary alone will not take you past grade 7; you need to practise explanatory writing.
备考建议：把 OCR 官方 specification 的每一行都当成一个可能的题干。看到一个知识点，先问自己三个问题：它是什么，它为什么这样设计，它在什么场景下会出错。这三个问题对应了 OCR 最常出的三档分值。
A practical revision rule: treat every line of the OCR specification as a possible question stem. For each knowledge point, ask three questions: what is it, why is it designed this way, and in what situation does it fail. These three questions map onto the three most common mark tariffs in OCR papers.
二、数据表示：二进制、十六进制与单位换算
数据表示是 Paper 1 的高频计算区，也是最好拿满分的部分。核心技能有四个：二进制与十进制的互相转换、十六进制转换、存储单位换算、以及图像与声音文件大小的估算。二进制转十进制用位权相加：10110110 的第 1、3、5、6、7 位是 1，对应 128 加 32 加 16 加 4 加 2，结果是 182。十进制转二进制用减权法：把 200 减去 128 得 72，减去 64 得 8，减去 8 得 0，所以在 128、64、8 三个位置写 1，其余写 0，得到 11001000。
Data representation is the highest-frequency calculation area in Paper 1 and the easiest place to secure full marks. Four skills matter most: binary to denary conversion, hexadecimal conversion, storage unit conversion, and estimating the file size of images and sound. To convert binary to denary, add the place values where a 1 appears: 10110110 has 1s in the 128, 32, 16, 4 and 2 positions, giving 182. To convert the other way, use the subtraction method: 200 minus 128 leaves 72, minus 64 leaves 8, minus 8 leaves 0, so we set the 128, 64 and 8 bits to 1 and all others to 0, producing 11001000.
十六进制常被考在意想不到的地方。记住一位十六进制等于四位二进制，所以 8 位二进制正好写成两位十六进制。二进制 11010101 分成 1101 和 0101，分别是 D 和 5，结果是 D5。存储单位要背准：1 KB 等于 1000 字节，1 KiB 等于 1024 字节，1 MB 等于 1000 KB，1 GiB 等于 1024 MiB。GCSE 阶段按 1000 进制的十进制前缀作答，除非题目明确要求二进制前缀。
Hexadecimal questions often appear in unexpected places. One hexadecimal digit equals four binary digits, so eight bits map neatly onto two hex digits. Split 11010101 into 1101 and 0101, which are D and 5, giving D5. Learn the storage units precisely: 1 KB is 1000 bytes, 1 KiB is 1024 bytes, 1 MB is 1000 KB, and 1 GiB is 1024 MiB. At GCSE, use decimal prefixes based on 1000 unless the question explicitly asks for binary prefixes.
图像文件大小的公式是 宽度 乘 高度 乘 色深，单位是比特，再除以 8 换算成字节。一张 400 乘 300 像素、色深 8 位的图片是 400 乘 300 乘 8 等于 960000 比特，也就是 120000 字节，约 120 KB。声音文件的公式是 采样率 乘 采样精度 乘 时长 乘 声道数。以 44100 Hz、16 位、60 秒、双声道为例：44100 乘 16 乘 60 乘 2 等于 84672000 比特，约 10.6 MB。这类题的失分点几乎永远是单位没换算，答题时把单位写在式子后面可以强制自己检查。
The file size formula for an image is width times height times colour depth in bits, divided by 8 to convert to bytes. A 400 by 300 pixel image with a colour depth of 8 bits is 400 times 300 times 8, which equals 960000 bits, or 120000 bytes, roughly 120 KB. The formula for sound is sample rate times bit depth times duration times number of channels. For 44100 Hz, 16 bits, 60 seconds and two channels: 44100 times 16 times 60 times 2 equals 84672000 bits, about 10.6 MB. Almost every lost mark here comes from forgetting the unit conversion, so write the unit after each figure so you are forced to check it.
三、算法：伪代码、流程图与搜索排序
Paper 2 的算法部分有三个考法：读伪代码写输出、补全伪代码、用自然语言或流程图描述算法。阅卷时是按步骤给分的，因此即使最终结果算错，只要中间步骤正确仍然能拿分。读伪代码时养成"逐行追踪变量表"的习惯：在草稿纸上画一张表，每执行一行就更新一次变量的值，循环体要写清楚每一轮的状态。这样既不容易算错，也方便复查。
The algorithm section of Paper 2 is examined in three ways: reading pseudocode and stating the output, completing pseudocode, and describing an algorithm in words or as a flowchart. Marks are awarded for method, so even an incorrect final answer can still earn most of the available marks if the intermediate steps are right. When tracing pseudocode, keep a variable table on paper and update it after every line, recording the state at the end of each loop iteration. This reduces arithmetic slips and makes checking far quicker.
线性搜索与二分搜索的区别是必考内容。线性搜索从第一个元素开始逐个比较，数据不需要排序，但平均比较次数是元素总数的一半，适合数据量小或无序的情况。二分搜索要求数据已排序，每次比较中间元素并排除一半，在 1000 个元素中最多比较 10 次左右，效率明显更高。答题时必须写出"数据必须是有序的"这个前提，否则会丢掉前提分。
The difference between a linear search and a binary search is examined every year. A linear search compares each item from the start, works on unsorted data, and needs on average half as many comparisons as there are items, which suits small or unordered data sets. A binary search requires the data to be sorted, compares the middle item and discards half the list each time, so a list of 1000 items needs at most about 10 comparisons. Always state the sorted prerequisite, otherwise you lose the precondition mark.
排序算法要掌握冒泡排序、插入排序与归并排序的思想与比较次数。冒泡排序反复比较相邻元素并交换，每一轮把最大的元素推到末尾，n 个元素最多需要 n 减 1 轮。插入排序把每个新元素插入到已排序序列的正确位置，在小规模或近乎有序的数据上比冒泡更快。归并排序把列表不断二分再合并，效率更高但需要额外空间。答题时不要只写"更快"，要写出"每一轮减少一个待排序元素"这样的过程描述。
For sorting, you need the ideas and comparison counts of bubble sort, insertion sort and merge sort. Bubble sort repeatedly compares adjacent items and swaps them, pushing the largest item to the end of each pass, so n items need at most n minus 1 passes. Insertion sort places each new item into the correct position in an already sorted sublist and beats bubble sort on small or nearly sorted data. Merge sort repeatedly splits the list and merges the halves, which is faster but needs extra memory. Do not answer with the word faster alone; describe the process, such as one fewer item to sort in each pass.
四、编程与伪代码书写规范
OCR 的伪代码有自己的规范，用缩进表示代码块，赋值用等号，输出用 print，输入用 input，条件语句用 if 和 else 加 endif，循环用 for 或 while 加 endwhile。不要直接写 Python 语法去替代要求书写的伪代码，除非题目允许。编程大题的常见结构是：读取输入、循环处理、条件判断、累加统计、输出结果。写之前先花 30 秒列一个变量清单，把每个变量的初值和类型写下来，可以显著减少逻辑漏洞。
OCR pseudocode has its own conventions: indentation marks code blocks, the equals sign assigns a value, print outputs, input reads a value, conditions use if and else closed by endif, and loops use for or while closed by endwhile. Do not substitute Python syntax when the question asks for pseudocode unless it explicitly allows it. The extended programming question usually follows a fixed shape: read input, loop, test conditions, accumulate totals, output results. Spend 30 seconds listing your variables with their initial values and types before writing; this removes most logic errors.
再举一个实际例子：统计班级分数并输出及格人数。变量清单是 total 初值 0、count 初值 0、score 输入值。伪代码写法是 for i 等于 1 到 30、input score、if score 大于等于 40 则 count 加 1、endif、endfor、print count。阅卷时会逐项检查：有没有初始化、循环次数是否正确、比较符号是大于等于还是大于、输出的是人数还是分数。写出完整骨架比写出漂亮代码更重要。
Take a concrete example: counting how many students in a class passed. The variable list is total starting at 0, count starting at 0, and score as input. In pseudocode: for i equals 1 to 30, input score, if score is greater than or equal to 40 then add 1 to count, endif, endfor, print count. Markers check each element: whether variables are initialised, whether the loop count is right, whether the comparison is greater than or greater than or equal to, and whether you output a count rather than a score. A complete skeleton beats elegant but incomplete code.
调试与测试也是考点。语法错误会让程序无法运行，逻辑错误会让程序运行但结果不对。测试时要设计正常数据、边界数据与错误数据三类：例如循环 1 到 30，边界测试要检查 1 和 30，错误测试要检查输入非数字时的行为。能写出"为什么选这组测试数据"的学生，往往在解释题上也能拿满分。
Debugging and testing are also examined. Syntax errors stop the program from running, while logic errors let it run but produce wrong output. Design three kinds of test data: normal, boundary, and erroneous. For a loop from 1 to 30, boundary tests check 1 and 30, and erroneous tests check what happens when a non-numeric value is entered. Students who can justify why they chose a set of test data usually write the strongest explanation answers too.
五、计算机系统、网络与网络安全
Paper 1 的系统部分要建立"层次感"。CPU 由控制单元、算术逻辑单元、寄存器与缓存构成，冯诺依曼架构的关键是"指令与数据存放在同一内存中，按取指、译码、执行循环运行"。内存分 RAM 与 ROM，RAM 断电丢失数据但可读写，ROM 断电保留数据但通常只读。存储介质比较时要落到具体指标：容量、速度、便携性、耐用性、可靠性、成本。只写"硬盘容量大"很难拿到满分，要写成"磁性硬盘每 GB 成本最低，适合存放大量备份数据，但含有移动部件，受撞击容易损坏"。
The systems section of Paper 1 rewards layered thinking. The CPU contains the control unit, the arithmetic logic unit, registers and cache, and the key idea of the Von Neumann architecture is that instructions and data share the same memory and are processed in a fetch, decode, execute cycle. Memory splits into RAM and ROM: RAM loses its contents when power is removed but is readable and writable, while ROM keeps its contents but is normally read only. When comparing storage media, use concrete criteria: capacity, speed, portability, durability, reliability and cost. Writing a hard drive has a large capacity rarely earns full marks; write that a magnetic hard drive has the lowest cost per gigabyte, suits large backups, but contains moving parts and is easily damaged by impact.
网络部分要分清硬件、拓扑与协议。硬件包括网卡、交换机、路由器、无线接入点；拓扑包括星型与网状；协议要能说出 HTTP、HTTPS、FTP、SMTP、IMAP、TCP 与 IP 各自负责什么。星型拓扑的优点是单个节点故障不影响其他节点，缺点是依赖中央交换机。互联网的分层思想也常考：数据被拆成数据包，每个包带有目标地址，经由路由器逐跳转发，到达后可能顺序错乱，再由 TCP 负责重组与确认。
For networks, separate hardware, topology and protocols. Hardware includes network interface cards, switches, routers and wireless access points; topologies include star and mesh; protocols include HTTP, HTTPS, FTP, SMTP, IMAP, TCP and IP, and you must be able to state what each one does. The star topology keeps one faulty node from affecting the others but depends on the central switch. Layering is also examined: data is split into packets, each packet carries a destination address, routers forward them hop by hop, they may arrive out of order, and TCP reassembles and acknowledges them.
网络安全是最好拿分也最容易吃亏的部分。答题要成对写：威胁与防护。恶意软件通过邮件附件传播，对应防护是杀毒软件与不打开可疑附件；网络钓鱼骗取凭证，对应防护是检查发件域名与启用双重验证；暴力破解反复尝试密码，对应防护是强密码策略与登录失败锁定；拒绝服务攻击用大量请求压垮服务器，对应防护是流量过滤与限流。只写"安装防火墙"这种泛化答案，通常只能拿一分。
Network security is the easiest area to score in and the easiest to under-answer. Write threats and defences in pairs. Malware spreads through email attachments, and the defence is anti-malware software plus a rule not to open unexpected attachments. Phishing steals credentials, and the defence is checking the sender domain and enabling two-factor authentication. Brute force repeatedly guesses passwords, and the defence is a strong password policy with account lockout. A denial of service attack floods a server with requests, and the defence is traffic filtering and rate limiting. Generic answers such as install a firewall usually earn a single mark.
六、冲刺阶段的提分策略
第一，把近五年的真题按题型而不是按年份刷。同一类解释题集中做，你会迅速看出 OCR 的标准答案结构：先给术语，再给原因，最后给例子或影响。第二，建立自己的错题术语表，把英文术语和中文解释成对记录，例如 packet switching 写成分组交换并注明每一跳独立选择路径。第三，限时训练时优先做 8 分大题，先写要点框架再补细节，平均每分钟拿一分是最低要求。
First, work through the last five years of past papers by question type rather than by year. Grouping explanation questions together quickly reveals the structure OCR mark schemes expect: state the term, give the reason, then add an example or a consequence. Second, keep a personal terminology log pairing English terms with Chinese explanations, for example packet switching recorded with the note that each hop chooses its own route. Third, in timed practice, attack the 8 mark question first: sketch the key points, then add detail, aiming for at least one mark per minute.
第四，重视命令词。State 只要一句结论，Describe 要写过程或特征，Explain 必须给出因果链，Compare 必须有双方并用比较词，Discuss 需要正反两面。很多学生把 Explain 当作 State 来写，一句话答完就交卷，分数自然上不去。第五，计算题必须写出单位与公式，即使算错，公式分和方法分依然有效。第六，考前一天不要做新题，只复盘术语表和错题，保证休息。
Fourth, respect the command words. State needs a single conclusion, describe needs a process or feature, explain needs a causal chain, compare needs both sides with comparative language, and discuss needs arguments for and against. Many students treat explain as state, write one sentence and move on, which caps their score. Fifth, always show units and formulas in calculation questions, because formula and method marks survive an arithmetic slip. Sixth, avoid new questions the day before the exam; review your terminology log and error list instead, and protect your sleep.
计算机科学是一门"理解加练习"的学科，理解考点结构的人能用很少的时间拿到大多数分数，而只靠刷题不理解结构的人会在解释题上反复失分。把上面五个板块的核心概念、公式和答题句式整理成自己的清单，配合真题限时训练，从 6 分提升到 8 分通常只需要三到四周的规律复习。
Computer Science rewards understanding plus practice. Students who understand the structure of the assessment earn most marks with relatively little time, while students who only grind questions keep losing marks on explanation answers. Turn the core concepts, formulas and answer patterns from these five areas into your own checklist, pair it with timed past paper practice, and moving from a grade 6 to a grade 8 typically takes only three to four weeks of consistent revision.
戴睿（dere）Neo优选系列 创意双触摸屏高性能笔记本电脑大学生AI学习商务办公设计炒股娱乐手提轻薄本32/1TB¥4586.0📱 去京东购买京东精选
需要一对一辅导？Need One-on-One Tutoring?
A-Level / GCSE / IB / KS2-KS3 全科辅导
16621398022 同微信
Visit aleveler.com
或截图后用「扫一扫 → 相册」选择本图