The Computer Science glossary

The rule that a term containing a variable is redundant when it is combined with that variable on its own: $A + A \cdot B = A$ and $A \cdot (A + B) = A$.

A class that cannot be instantiated, used as a superclass to share code and to declare methods that its subclasses must provide.

A description of a collection of data in terms of the values it can hold and the operations that can be performed on it, independent of how the data is stored or how the operations are coded.

A method declared without a body in an abstract class or interface, which every concrete subclass must override with an implementation.

Focusing on the features of a problem that are essential for a given purpose and removing unnecessary detail, to produce a simpler model or overview of the problem.

The principle of exposing what an object can do through a set of method declarations while hiding how each class carries out those methods.

Honest and responsible conduct in academic work, in which submitted work is authentic and the student's own, and the ideas and work of others are fully acknowledged.

Testing of the complete product by its client or intended users, carrying out real tasks, to decide whether it meets the agreed requirements.

The security function that decides which users and programs may use which resources, and in what way (read, write or execute), according to rules set by the owner or administrator.

A keyword that sets which parts of a program may use a class member, such as private (only code in the same class) or public (any code).

A public method that returns the value of an object's attribute without changing it, usually named getX() (Java) or get_x() (Python).

The obligation of the people and organisations who build and deploy a system to take responsibility for its decisions and their consequences, to explain them, and to put right any harm caused.

The operating system function that records how much of each resource (CPU time, memory, storage, network data) each user and process uses, for monitoring, quotas and billing.

A register that holds the result of the most recent ALU operation and supplies one of the inputs to the next.

A variable that builds up a result, such as a running total or product, by being updated with each value processed inside a loop.

The proportion of all predictions that a classifier gets right: (TP + TN) divided by the total number of examples.

The four properties that a DBMS guarantees for every transaction so that transactions are processed reliably: atomicity, consistency, isolation and durability.

A message from the receiver confirming that data has arrived, which in TCP gives the sequence number of the next byte the receiver expects.

One of the choices available to an agent in a state, which changes the environment and leads to a reward and a new state.

The function a neuron applies to its weighted sum to produce its output; non-linear activation functions let networks model non-linear patterns.

An output device that converts a control signal into physical action.

The bus that carries memory addresses from the CPU (from the MAR) to main memory, in one direction only.

The symmetric block cipher standardised by the US National Institute of Standards and Technology (NIST) in 2001, which encrypts data in 128-bit blocks using a key of 128, 192 or 256 bits.

An input deliberately altered, often by changes too small for a person to notice, so that a machine learning model gives a wrong output.

A technique that prevents starvation by gradually raising the priority of a process the longer it waits in the ready queue.

In reinforcement learning, the program that observes its environment, chooses actions and learns from the rewards those actions earn.

An SQL function that takes the values of a column from many rows (a whole table or one group) and returns a single summary value; the standard ones are AVG, COUNT, MAX, MIN and SUM.

A weak has-a relationship in which the containing object holds references to component objects that are created outside it, can be shared or moved, and continue to exist when it is destroyed.

An approach to software development in which a product is built in a series of short cycles, each delivering a working, tested part of the product, with the requirements and the plan reviewed with the client between cycles.

Compiling a program fully into native machine code for a target platform before it is run, so that no translation is needed while it executes.

A precise, finite sequence of well-defined instructions that solves a problem or performs a task.

A systematic and repeatable error in a computer system that produces unfair outcomes, such as favouring or disadvantaging one group of people over another.

The process of developing a step-by-step series of instructions that solves a problem, and refining it until it is correct and efficient.

A measure of the amount of computing resources, mainly time and memory, that an algorithm uses as the size of its input increases.

The property of an automated decision system whose outcomes or error rates do not unjustly differ between groups of people, measured against a stated definition of fairness.

A formula that gives the maximum speed-up of a task when part of it is run in parallel on several processors, showing that the part that must run sequentially limits the overall speed-up.

A signal that varies continuously, so it can take any value within a range at any moment in time.

A circuit that measures an analogue signal at regular intervals and outputs each measurement as a binary number.

A logic gate whose output is 1 only when all of its inputs are 1.

A data mining technique that identifies records, events or observations that differ significantly from the normal pattern of the data, such as fraudulent transactions or faulty sensor readings.

The condition of acting or communicating without one's identity being known; pseudonymity is acting under a consistent false name that is not linked to one's real identity.

The "if" part of an association rule, the itemset on the left of the arrow.

Data stored so that new records are only ever added to the end of the existing data, and existing records are never modified or removed.

The top layer of the TCP/IP model, which holds the protocols that programs use to exchange data for a particular service.

A documented set of functions, rules and data formats that one piece of software offers so that other software can use its services without knowing how they are implemented.

Software that performs a specific task for the user, rather than managing the computer itself.

A chip designed and manufactured to perform one specific task, giving the highest speed and energy efficiency for that task but unable to be changed for other tasks.

An algorithm for association rule mining that finds frequent itemsets level by level (single items, then pairs, then triples), counting only candidate itemsets whose subsets are all frequent, then builds rules from them that meet a minimum confidence.

A value or expression supplied in a subprogram call, which is copied into the matching parameter.

The part of the CPU that performs arithmetic operations, such as addition and subtraction, and logical operations, such as AND, OR, NOT and comparisons.

A symbol that performs a mathematical calculation on numeric values and produces a numeric result.

A fixed-size, ordered collection of elements of the same data type, stored contiguously in memory and accessed by an integer index starting at 0.

A Java class that stores an ordered, indexed collection of objects in an internal array, which it replaces with a larger one automatically when it fills.

The field of computer science that builds systems able to perform tasks normally thought to need human intelligence, such as recognising speech, making decisions or playing games.

The European Union regulation, in force from 1 August 2024 with its provisions phased in over the following 6 to 36 months, that regulates AI systems according to their level of risk.

A machine learning model made of layers of connected units (artificial neurons), in which each unit combines its weighted inputs and passes the result to the next layer; training adjusts the weights.

The American Standard Code for Information Interchange, a 7-bit character set of 128 codes covering the English alphabet, digits, punctuation and control characters.

One of the skills the IB assesses in a course, numbered from AO1 (knowledge and understanding) through AO2 (applying knowledge) to AO3 (constructing, analysing and evaluating).

A statement that evaluates an expression and stores the result in a variable, replacing any value the variable held before.

A relationship between two classes in which objects of one class are connected to, and can use, objects of the other, drawn in UML as a solid line between the classes.

A rule of the form X → Y ("if X, then Y") stating that when the items or events in X occur in a record, the items or events in Y also tend to occur, together with measures of how often and how reliably this holds.

The rule that the grouping of three or more operands joined by the same operator does not change the result: $(A \cdot B) \cdot C = A \cdot (B \cdot C)$ and $(A + B) + C = A + (B + C)$.

Encryption that uses a mathematically related pair of keys, in which data encrypted with one key of the pair can only be decrypted with the other.

A value that cannot be split into smaller parts that the database needs to use separately, so that each field of each record holds exactly one such value.

The property that a transaction is all or nothing: either every one of its changes is applied, or, if any part fails, none of them is.

The gradual loss of strength of a signal as it travels through a transmission medium, which limits how far it can go before it must be amplified or regenerated.

A named item of data that a class declares and each object holds a value for.

Technology that overlays computer-generated images, sound or information onto a person's view of the real world in real time, through a phone screen, headset or smart glasses.

The process of checking the identity of a user, device or program before it is given access.

The expected number of steps an algorithm performs over all possible inputs of size $n$, assuming each input is equally likely, and the complexity this gives.

A self-balancing search tree in which each node holds many sorted keys and has many children, keeping the tree shallow so that few nodes need to be read from disk during a search.

The algorithm that calculates how much each weight in a neural network contributed to the loss, by passing the error backwards from the output layer through the hidden layers.

A tree in which, at every node, the heights of the left and right subtrees differ by at most a small fixed amount (at most one in the strictest definition), so that the height stays close to $\log_{2} n$.

The maximum amount of data that a link can carry per second, measured in bits per second.

A control algorithm in which the actuator is switched fully on or fully off depending on whether the measured value is below or above the set point, often with hysteresis (separate on and off thresholds) to stop rapid switching.

A consistency model used by many distributed NoSQL databases that puts availability before immediate consistency. Its letters stand for basic availability (every request gets an answer, even when some copies cannot be reached), soft state (copies may change without new input while updates spread between them) and eventual consistency.

The condition in a recursive subprogram under which it returns a result directly, without calling itself again.

The input of a given size $n$ for which an algorithm performs the fewest steps, and the complexity on that input.

A way of marking against level descriptors in which the marker chooses the descriptor that most accurately matches the work overall, so a strength in one aspect can compensate for a weakness in another.

A learnable number added to a neuron's weighted sum, which shifts the threshold at which the neuron activates, independently of its inputs.

A systematic test of an automated decision system, often by an independent party, that compares its outputs and error rates across groups of people to find and measure unfair differences.

Data sets so large, so fast-growing or so varied in form that they cannot be stored and processed efficiently with traditional single-server database tools.

A way of describing the time or space complexity of an algorithm by the term that grows fastest as the input size $n$ increases, ignoring constant factors and smaller terms, usually for the worst case.

The base-2 number system, which uses only the digits 0 and 1 and column values that are powers of 2 (1, 2, 4, 8, 16, …).

A data type that stores a large block of raw binary data whose contents the DBMS does not interpret.

A unit prefix based on powers of 1024, as in kibibyte (KiB, 1024 bytes), mebibyte (MiB, $1024^{2}$ bytes) and gibibyte (GiB, $1024^{3}$ bytes).

A search algorithm for a sorted list that compares the target with the middle element of the part still being searched and discards the half that cannot contain it, repeating until the target is found or no elements remain.

A binary tree in which, for every node, every value in its left subtree is less than the node's value and every value in its right subtree is greater.

A tree in which every node has at most two children, identified as the left child and the right child.

A binary digit, the smallest unit of data, which has one of two values, 0 or 1.

The number of bits used to store each sample of a sound, which sets how many different amplitude levels ($2^{n}$ for $n$ bits) can be recorded.

An image stored as a grid of pixels, with a binary value recording the colour of each pixel.

A system whose inputs and outputs can be observed but whose internal reasoning cannot practically be inspected or understood, either because it is too complex or because it is kept secret.

A list of items, such as IP addresses, ports, websites or devices, that are explicitly blocked, with everything not on the list allowed by default.

A distributed, append-only record of transactions, grouped into blocks that each contain a cryptographic hash of the previous block and are copied across many nodes in a peer-to-peer network.

A data type with exactly two values, true and false, used to store the result of a condition.

The algebra of the two values true (1) and false (0), with operations such as AND, OR and NOT and laws for rewriting and simplifying expressions.

An expression built from Boolean variables, the constants 0 and 1, and Boolean operators, whose value is either true (1) or false (0).

An operation that combines or changes Boolean values: AND, OR, NOT, NAND, NOR, XOR or XNOR.

The firmware program, stored in ROM, that runs when a computer is switched on; it checks the hardware, then loads the operating system from storage into RAM and starts it.

A network of devices infected with malware that an attacker controls remotely, usually through a command-and-control server, without the owners' knowledge.

Test data at the limits of the valid range, and just outside them, used to check that the program accepts and rejects values at exactly the right point.

A diagram of one numerical variable that shows its median, lower and upper quartiles as a box, the spread of the remaining non-outlying values as whiskers, and any outliers as separate points.

Recursion in which a call of the subprogram can make two or more recursive calls, so the calls form a tree.

A statement that ends the innermost loop immediately and continues with the first statement after the loop; the related continue statement skips the rest of the current pass and starts the next one.

A marker placed on a line of code that makes the debugger pause the program just before that line runs.

The set of devices that receive a broadcast sent by any device in the set; a router marks its boundary.

An attack that tries every possible key or password in turn until the correct one is found.

A sorting algorithm that makes repeated passes through a list, comparing each pair of adjacent elements and swapping them if they are in the wrong order, until a pass makes no swaps.

One position in a hash table's array, identified by an index, which holds the key or keys whose hash values lead to it.

An area of main memory that holds data temporarily while it is moved between a program and a slower device such as secondary storage.

A function supplied ready-made by a programming language or its standard library, which a program can call without writing its algorithm.

A set of parallel wires that carries signals between components of a computer, one bit per wire at a time.

A topology in which all nodes share a single cable, the backbone, so every transmission reaches every node.

The technologies and processes that collect, analyse and present an organisation's data, through reports, dashboards, OLAP and data mining, to support decision-making.

A group of 8 bits, the standard unit for measuring storage and for addressing memory.

A compact, platform-independent set of instructions produced by compiling source code, designed to be executed by a virtual machine rather than directly by a CPU.

The part of a virtual machine that executes bytecode one instruction at a time by fetching each instruction, decoding it and running the matching routine, without converting it into native machine code.

A small amount of very fast memory, inside or next to the CPU core, that holds copies of the instructions and data from RAM that are most likely to be needed next.

A memory access in which the requested instruction or data is found in the cache, so it is supplied quickly without reading RAM.

A memory access in which the requested instruction or data is not in the cache, so it must be fetched from a slower level (a lower cache level or RAM) and is usually copied into the cache for next time.

A stack maintained by the running program that holds one entry for each subprogram call that has started but not yet returned, with the most recent call on top.

A field, or a minimal set of fields, whose values are unique for every record in a table, so that it could be chosen as the primary key.

The result that a distributed data store cannot guarantee all three of consistency (every read sees the latest write), availability (every request receives a response) and partition tolerance (the system continues when network messages between nodes are lost); when a network partition occurs, it must give up either consistency or availability.

The number of elements a dynamic array can hold in its current block of memory before it must allocate a larger block.

The maximum number of instances of one entity that can be related to a single instance of another entity in a relationship, recorded as one or many.

The result of combining two tables so that every row of the first is paired with every row of the second, giving (rows in first) × (rows in second) rows.

A booklet published by the IB before the examination that describes a computing scenario and its challenges, which students research in advance and are then examined on in Paper 1 Section B.

The component of a computer that fetches, decodes and executes the instructions of a program, processing data and coordinating the other hardware.

The centre of a cluster, calculated as the mean of each feature over all the points in the cluster.

A trusted organisation that checks that an applicant controls a domain or identity and then issues digital certificates signed with its own private key.

The sequence of certificates from a website's certificate through any intermediate certificates to a root certificate, in which each certificate is signed by the one above it and the root is already trusted by the device.

A data type that holds exactly one character, stored as that character's numeric code.

The number that a character set assigns to one character, stored in binary to represent that character.

An agreed list of characters in which each character is given a unique number, so that text can be stored and exchanged as binary.

In Java, an exception that the compiler requires a method either to catch or to declare with throws, used for problems outside the program's control.

A value calculated from a block of data and sent with it, so that the receiver can repeat the calculation and detect whether the data was corrupted in transit.

A node directly below another node and linked to it by an edge; every node except the root is the child of exactly one parent.

In a genetic algorithm, the encoding of one candidate solution.

A way of writing an IP network as an address followed by a slash and the number of bits in its network part.

The unreadable data produced by encrypting plaintext.

A method of communication in which a dedicated path with a fixed capacity is set up between two devices before any data is sent and reserved for them until the communication ends.

A linked list in which the last node's reference points back to the first node, so the nodes form a closed loop with no null reference.

A queue stored in a fixed-size array in which the front and rear pointers wrap round from the last index to index 0, so slots freed by dequeuing are reused.

A short acknowledgement at the point where a source is used, which points the reader to a full entry for that source in a bibliography or reference list.

A template that defines the attributes and methods shared by every object of one kind.

The tree of classes formed by inheritance, with the most general superclass at the root and increasingly specialised subclasses below it.

A situation in which one class is far more common than another in a data set, so that a model can score high accuracy while performing badly on the rare class.

A supervised learning task in which the model predicts which of a fixed set of categories (classes) an input belongs to.

One part of an SQL statement, starting with a keyword such as SELECT, FROM, WHERE or ORDER BY, that does one job in the statement.

A device, or a program running on one, that requests a service or resource from a server.

The person or organisation who has the problem an IA product is built to solve, and who provides requirements, feedback and help with testing.

A networking model in which central servers store resources and provide services, and clients send requests to the servers and receive responses.

A circuit that produces a steady stream of electrical pulses to synchronise the operations of the CPU; its rate, measured in hertz, is the number of pulses (cycles) per second.

A control system that measures its output with sensors and feeds the measurement back to the controller, which compares it with the desired value and adjusts the output to reduce the difference.

The delivery of computing resources, such as servers, storage, databases, software and processing power, over the internet on demand, with the provider owning and maintaining the hardware and the customer paying according to use.

A database hosted and run by a cloud provider and accessed over the internet, often as a managed service in which the provider handles installation, maintenance, backups and scaling.

A service from a cloud provider that rents out machine learning hardware such as GPUs and TPUs on demand, together with storage and tools for training, deploying and monitoring models.

An unsupervised learning task that divides unlabelled examples into groups (clusters) so that examples in the same group are more similar to each other than to examples in other groups.

A processor that supplements the CPU by carrying out one specific type of task, so that the CPU is free for other work.

A measure of how well a regression model fits its data, equal to the proportion of the variation in the dependent variable that the model explains, from 0 (none) to 1 (all).

The four conditions that must all hold at once for deadlock to occur: mutual exclusion, hold and wait, no pre-emption and circular wait.

A recommendation technique that predicts a user's interest in an item from the ratings or behaviour of other users with similar tastes.

The part of a network in which a transmission from one device can collide with a transmission from another.

The number of bits used to store the colour of each pixel; a colour depth of $n$ bits allows $2^{n}$ different colours.

A temporary name given to a column of the result set, or to a table within one query, using the keyword AS.

The instruction word at the start of an examination question that tells the student what kind of answer is required and how much depth it needs.

A user interface in which the user types text commands and the system replies with text output.

The TCL command that ends a transaction successfully, making all its changes permanent and visible to other transactions.

The rule that the order of the operands of AND or OR does not change the result: $A \cdot B = B \cdot A$ and $A + B = B + A$.

A translator that converts the whole of a program's source code into machine code (or another lower-level form) in one process before the program is executed, reporting the errors it finds in the whole program.

The rule that ANDing a variable with its inverse gives 0, and ORing a variable with its inverse gives 1: $A \cdot \overline{A} = 0$ and $A + \overline{A} = 1$.

A primary key made of two or more fields, whose combined values are unique for each record even though each field on its own may repeat.

A strong has-a relationship in which the containing object creates and owns its component objects, which cannot exist independently of it and are destroyed with it.

Encoding data using fewer bits than the original representation, to reduce the storage space it needs or the time it takes to transmit.

The size of the original data divided by the size of the compressed data, usually written as a ratio such as 10 : 1.

The specific area of computing selected for an IA solution, including the kind of product and, usually, the main languages and tools used to build it.

The thought processes involved in formulating a problem and expressing its solution so that a computer, whether a human or a machine, can carry it out effectively.

Two or more computing devices connected by a transmission medium so that they can exchange data and share resources.

A key whose value is formed by joining the values of two or more fields into a single value.

Joining two or more strings end to end to make a new string.

A high-level description of the data an organisation needs, as entities, their attributes and the relationships between them, independent of any database management system or storage method.

A class that implements all of its methods, including every abstract method it inherits, and so can be instantiated.

The methods a DBMS uses to manage transactions that run at the same time so that they do not interfere with each other and the result is the same as if they had run one after another.

A Boolean expression, evaluated while the program runs, whose true or false result controls a selection or a loop.

A loop that repeats its body for as long as a condition remains true (or until it becomes true), so the number of repetitions is not known when the loop starts.

For a rule X → Y, the proportion of the transactions containing X that also contain Y: support of X and Y together divided by support of X.

A table that counts a classifier's predictions against the actual classes, so that each cell shows how many examples of one actual class were predicted as each class.

The condition in which more traffic is sent across a link or device than it can carry, causing delays and lost packets.

A small circle containing a label (a letter or number) that joins two parts of a flowchart without drawing a flowline between them; the flow continues from one connector to the other connector with the same label.

A person's free, specific and informed agreement to their data being collected and used for a stated purpose, which they can also withdraw.

The "then" part of an association rule, the itemset on the right of the arrow, which tends to occur when the antecedent does.

The property that a transaction takes the database from one valid state to another, so that every constraint and rule holds after the transaction commits.

A named value that is set once and cannot be changed while the program runs.

A complexity in which the number of steps stays the same however large the input becomes.

A restriction that a solution must work within, such as a limit on time, budget, hardware, software, law or the skills of its users.

A special method that runs automatically when an object is created and sets the initial values of the object's attributes.

A group of servers in many locations that store copies of content and deliver each request from the server nearest the user.

The process by which an online platform reviews user-generated content and removes, restricts or labels material that breaks its rules or the law, using automated classifiers, human moderators or both.

A set of rules a website sends to the browser in an HTTP header, stating which sources of scripts, styles and other content the browser may load and run for that site.

The act of saving the state of the process currently using the CPU and loading the saved state of another process, so that the CPU can be passed between processes without either losing its place.

The set of rules, implemented in software, that a controller uses to calculate its outputs from the sensor inputs and the set point.

The bus that carries control and timing signals between the CPU and other components, such as memory read, memory write, clock pulses and interrupt requests.

A system of hardware and software that manages the behaviour of a device or process by processing inputs and producing outputs that act on it, often using feedback to keep a measured quantity close to a target value.

The part of the CPU that decodes each instruction and sends control signals that coordinate the other components, directing the flow of data between the registers, the ALU, memory and input/output devices.

The component of a control system that receives data from sensors, compares it with the set point using a control algorithm, and sends output signals to the actuators.

The point at which a repeated learning process stops changing its result.

The state in which, after a change in the network, every router has received the updates and recalculated its routes, so that all routing tables agree on the new topology; convergence time is how long this takes.

A layer of a CNN that slides each of its kernels across its input, calculating a weighted sum at every position, to produce one feature map per kernel.

A deep neural network designed for images and other grid-like data, which uses convolutional layers of small learnable kernels and pooling layers to learn features from local regions, followed by fully connected layers that classify.

The build-up of short processes waiting behind one long process under first-come, first-served scheduling, which raises the average waiting time and leaves devices idle while the long process holds the CPU.

An independent processing unit within a processor, with its own control unit, ALU and registers, able to fetch, decode and execute its own stream of instructions.

A number from −1 to +1 that measures the strength and direction of the straight-line relationship between two variables.

A loop that repeats its body a number of times fixed before the loop starts, using a loop variable that steps through a range of values.

A variable that counts how many times something happens, usually starting at 0 and increased by 1 each time the event occurs.

The degree to which one class depends on the internal details of another, so that a change to one forces changes to the other.

Describes software that runs on several operating systems or types of hardware from the same source code or the same distributed file.

An attack in which a website includes untrusted input in a page without encoding it, so that a script supplied by the attacker runs in other users' browsers with the trusted site's permissions.

A method of evaluating a model in which the data is split into k equal parts (folds) and the model is trained k times, each time on k − 1 folds and validated on the remaining fold, with the k scores averaged.

In a genetic algorithm, the operation that combines parts of two parent chromosomes to produce one or more child chromosomes.

Interference in which the signal carried by one wire or pair of wires is picked up by an adjacent wire or pair in the same cable.

A way of drawing ERDs in which entities are boxes, relationships are lines between them, and a pair of symbols at each end of a line shows the cardinality and modality at that end.

An algorithm that turns input of any length into a fixed-length value (a digest) in such a way that the same input always gives the same digest, any change to the input gives a completely different digest, and the input cannot practically be worked out from the digest.

The total reward an agent collects over a sequence of actions, which it tries to maximise.

The collection of problems that arise as the number of features grows, including sparse data, the need for exponentially more records, distances between records becoming similar, overfitting, and rising computation and memory costs.

Dividing a customer base into groups whose members behave similarly, so that each group can be targeted differently.

The bus that carries data and instructions between the CPU (through the MDR) and main memory, in both directions.

The process of detecting and correcting or removing errors, inconsistencies, duplicates, irrelevant entries and gaps in a dataset so that it accurately represents what it describes.

The corruption of two transmissions that are sent at the same time on a shared medium, so that both must be sent again.

The state in which every copy, view and use of a piece of data shows the same value, so that no two parts of the database contradict each other.

The part of SQL whose statements create, change and remove the structure of a database, such as tables, views and indexes, rather than the data in them.

A table that documents every field in a database, giving its name, data type, size or format, key status, whether it may be left empty, its validation rules and a description of what it holds.

The same value or set of values stored more than once, in the same table or in different tables or files.

A diagram that shows how data moves through a system, using symbols for external entities, processes, data stores and labelled arrows for the data flows between them.

The ability to change the schema at one level without having to change the schema, or the programs that use it, at the next level up.

The accuracy, validity and reliability of data over its whole life, maintained by rules that the data must obey whenever it is entered or changed.

The use, during training or preprocessing, of information that would not be available when the model is used for real, which makes the model's measured performance unrealistically high.

The part of SQL whose statements read, add, change and remove the data stored in the tables, leaving the structure unchanged.

A smaller data warehouse, or a subset of one, focused on a single subject or department, such as sales, finance or marketing.

The principle that an organisation should collect and keep only the personal data that is adequate, relevant and necessary for its stated purpose.

The automatic discovery of useful, previously unknown patterns, relationships and anomalies in large data sets, using statistical and machine learning techniques.

Dividing a database's data into separate parts stored on different nodes, either by rows (horizontal partitioning, also called sharding) or by columns (vertical partitioning).

An attack in which false or manipulated examples are added to a model's training data so that the trained model behaves wrongly, either in general or for inputs chosen by the attacker.

The set of steps that turn raw data into a form a machine learning algorithm can learn from, including cleaning, encoding, scaling and selecting features.

The degree to which a dataset is accurate, complete, consistent, valid, unique, relevant and up to date for the purpose it is used for.

Data that is unnecessary because the same information is already stored elsewhere in the database or can be derived from other stored data.

The principle that data is subject to the laws of the country in which it is physically stored or processed.

The condition in which most of the possible combinations of feature values have no records, either because the space is too large for the data to fill or because most values in the dataset are empty or zero.

A way of organising and storing a collection of data items in memory, together with the operations used to access and change them.

A classification of data that sets which values a variable can hold, how they are stored and which operations can be performed on them.

A central database that stores integrated, historical data copied from many operational systems, organised by subject and optimised for analysis and reporting rather than for day-to-day transactions.

An organised, persistent collection of related data, stored so that it can be searched, updated and shared by many users and programs.

A separate data structure, kept by the DBMS, that stores the values of one or more columns in sorted order with a pointer to each matching row, so that rows can be found without reading the whole table.

The software that creates and maintains a database and controls every access to it, handling storage, retrieval, integrity rules, concurrent use, security and recovery.

A formal description of the structure of a database: its tables, the fields in each table, their data types, the keys and relationships that link the tables, and the constraints the data must obey.

A structured collection of examples used to train or test a machine learning model, where each example records the same set of properties.

Data types that store calendar dates (DATE), times of day (TIME) or both together (DATETIME or TIMESTAMP) in a standard form that the DBMS can validate, compare, sort and use in date arithmetic.

Techniques and services that keep a service available during a DDoS attack, such as rate limiting, filtering traffic through a scrubbing service and spreading load across a large distributed network.

The two rules $\overline{A \cdot B} = \overline{A} + \overline{B}$ and $\overline{A + B} = \overline{A} \cdot \overline{B}$, which state that the inverse of an AND is the OR of the inverses, and the inverse of an OR is the AND of the inverses.

A situation in which two or more processes are each waiting for a resource held by another process in the same group, so that none of them can ever continue.

A tool, usually part of an IDE, that runs a program under the programmer's control, pausing it and showing the values of its variables so that errors can be located.

The process of finding, diagnosing and correcting errors in a program.

A unit prefix based on powers of 1000, as in kilobyte (kB, 1000 bytes), megabyte (MB, $1000^{2}$ bytes), gigabyte (GB, $1000^{3}$ bytes) and terabyte (TB, $1000^{4}$ bytes).

The diamond-shaped flowchart symbol for a condition that is either true or false, with a separate labelled flowline leaving for each answer.

A supervised learning model that classifies data by following a path from a root node through tests on feature values, one test per internal node, to a leaf node that gives the predicted class.

A statement that introduces a variable to a program, giving its identifier and, in statically typed languages such as Java, its data type.

A programming language in which the programmer states what result is wanted, and the system works out how to produce it.

Breaking a complex problem or system into smaller, more manageable sub-problems that can each be understood, solved and tested separately.

A tree diagram that shows a problem at the top and, on each lower level, the sub-problems it is broken into.

Machine learning that uses artificial neural networks with many layers, which learn their own features in stages directly from raw data such as images, audio or text.

An image, video or audio recording of a real person that has been created or altered using deep learning so that it appears authentic.

The level of access Java gives a member declared with no access modifier, which lets it be used only by classes in the same package.

A security strategy that uses several independent layers of protection, so that an attack that defeats one layer is stopped or detected by another.

A tree in which every parent node has only one child, so the tree has the shape and performance of a linked list.

Handling missing data by removing every record that contains a missing value, or by removing a feature whose values are mostly missing.

A fault in a poorly designed table where deleting a record unintentionally removes the only copy of another fact.

A character or sequence of characters that marks the boundary between fields in a line of text, such as a comma, a tab or a colon.

The base-10 number system, which uses the ten digits 0 to 9 and column values that are powers of 10.

An attack that makes a service unavailable to its legitimate users by overloading it with traffic or requests, or by exploiting a flaw that makes it crash.

The deliberate introduction of redundancy into a normalised database, by combining tables or storing copies or precomputed values, so that data can be read faster or with simpler queries.

The output variable in a model, whose value is predicted from the independent variable or variables; also called the response.

A double-ended queue, a data structure that allows items to be added and removed efficiently at both its front and its rear.

The queue operation that removes the item at the front of a queue and returns it.

A value that can be calculated from other data already stored in the database.

A general, reusable description of how to arrange classes and objects to solve a problem that recurs in software design.

A program that lets the operating system communicate with one particular type of hardware device by translating general requests into the commands that the device understands.

The period for which a DHCP server allows a client to use an assigned IP address before the client must renew it or give it up.

A server that automatically assigns IP addresses and other network settings to devices from a pool of addresses, each for a limited lease period.

The set of elements that are in the first set but not in the second; written $A - B$ or $A \setminus B$.

An electronic document, digitally signed by a certificate authority, that binds a public key to the identity of its owner, such as a website's domain name, for a stated period.

The gap between people, groups or countries that have access to modern digital technology, connectivity and the skills to use them and those that do not.

The shared networks, data centres, services and connected devices that other systems and organisations rely on to store, process and transmit data.

A signal that can take only a limited set of discrete values, such as the binary values 0 and 1, at discrete moments in time.

A value created by processing a digest of a message with the sender's private key, which anyone holding the sender's public key can check to confirm who sent the message and that it has not been changed.

A circuit that turns a sequence of binary numbers into a continuously varying analogue signal.

A table in a star schema that describes one aspect of the facts, such as date, product, store or customer, with descriptive attributes used to filter and group the facts.

The number of features used to describe each record in a dataset, which is the number of dimensions of the space its records lie in.

Reducing the number of features in a dataset while preserving the information that matters for the task, by selecting a subset of features or by combining features into fewer new ones.

A named container in a file system that holds files and other directories, so that files can be organised in a hierarchy.

A mathematical transform that expresses a block of values, such as 8 × 8 pixels, as a sum of cosine waves of different frequencies, giving one coefficient for each frequency.

False information that is created or spread deliberately to deceive, for example to influence an election, damage a reputation or make money.

A rule for calculating how far apart two records are from their feature values, such as Euclidean or Manhattan distance, used by algorithms that compare records by similarity.

A database whose data is stored across several computers (nodes) at one or more sites, connected by a network, but which appears to its users as a single database.

A denial-of-service attack in which the traffic comes from many devices at once, usually the members of a botnet, so that it cannot be stopped by blocking a single source.

A group of networked computers that coordinate their actions by passing messages, so that together they appear to users as a single system.

The property of a distributed database that hides from users and applications the fact that data is spread across several sites, so they work with it as if it were a single, local database.

The rule that an operator applied to a bracket can be applied to each term inside it: $A \cdot (B + C) = A \cdot B + A \cdot C$ and $A + B \cdot C = (A + B) \cdot (A + C)$.

An algorithm design strategy that divides a problem into smaller sub-problems of the same type, solves each sub-problem recursively, and combines their solutions to solve the original problem.

A server that receives a client's DNS query and finds the answer, from its cache or by querying root, top-level domain and authoritative servers in turn, before returning it to the client.

A NoSQL database that stores each record as a self-describing document, such as a JSON document, whose fields can be queried and indexed and can differ from one document to the next.

The rule that every value in a field belongs to the field's allowed set of values (its domain), defined by its data type, length, format and range.

A human-readable name that identifies a server or group of servers on the internet and is translated into an IP address before data can be sent.

The hierarchical, distributed system of servers that translates domain names into the IP addresses that computers use to find each other.

The rule that inverting a variable twice gives the original variable: $\overline{\overline{A}} = A$.

A linked list in which each node has two references, one to the next node and one to the previous node, so the list can be traversed in both directions.

An OLAP operation that moves from summarised data to more detailed data along a dimension's hierarchy, such as from year to quarter to month; the reverse operation, which summarises, is called roll-up.

A record that describes the same real-world entity or event as another record in the dataset, either exactly or with small differences in spelling or formatting.

The property that once a transaction has been committed, its changes are permanent and survive a later crash or power failure.

A data structure whose size can change while the program runs, with memory allocated and released as items are added and removed.

An application-layer protocol that automatically gives a device joining a network its IP address, subnet mask, default gateway and DNS server address, taken from a pool managed by a DHCP server.

An IP address that is assigned automatically by a DHCP server for a limited lease period and may change when the lease ends or the device reconnects.

Polymorphism in which the method to run is chosen while the program runs, from the actual class of the object that receives the call, through method overriding.

A method of routing in which routers exchange information about the networks they can reach and use it to build and update their routing tables automatically, so that routes change when the network changes.

A language feature in which a variable's data type is decided by the value it currently holds and checked only while the program runs.

A link between a parent node and one of its children.

Processing data on or near the device that produces it, at the "edge" of the network, in place of sending it all to a distant cloud data centre.

A device at the edge of a network, close to where data is produced, such as a phone, camera, sensor or car, that runs machine learning models locally instead of sending data to a remote server.

A way of choosing the number of clusters k by plotting the within-cluster sum of squares against k and picking the value at the "elbow", where further increases in k give only small reductions.

Unwanted electrical noise induced in a transmission medium by nearby electrical equipment or radio signals, which can corrupt the data being carried.

Discarded electrical and electronic equipment, such as computers, phones and servers, which contains valuable and toxic materials and is often exported to countries with weaker recycling controls.

A single value stored in an array or list, identified by its index.

A selection structure that tests a series of conditions in order and runs only the block belonging to the first condition that is true, with an optional final else block for when none is true.

Scanning incoming email for spam, phishing and malware, and blocking, quarantining or labelling suspicious messages before they reach users.

A feature selection strategy in which the learning algorithm itself selects or down-weights features while the model is being trained.

Flash storage packaged with its controller in a single chip that is soldered permanently onto a device's circuit board.

A computer system with a dedicated function that is built into a larger mechanical or electronic device, usually with limited memory, processing power and user interface.

The bundling of an object's data and the methods that operate on that data into a single unit, the class.

The process in which each layer of the TCP/IP model, on the sending host, wraps the data from the layer above with its own header (and sometimes a trailer), with the reverse process, de-encapsulation, removing each layer's header on the receiving host.

A network protocol that encrypts the data it carries and authenticates the other party, so that intercepted traffic cannot be read or altered undetected.

The process of converting readable data into an unreadable form using an algorithm and a key, so that only someone with the correct key can convert it back (decryption).

A secret value, usually a long string of bits, that controls how an encryption algorithm transforms plaintext into ciphertext and back.

The condition reached when the file pointer has passed the last character of a file, so that no data is left to read.

The queue operation that adds a new item at the rear of a queue.

A model that combines the predictions of several different models in order to predict more accurately than any of them alone.

A person, object, place, event or concept about which data is stored.

The rule that every record has a primary key value that is not null and is unique within its table, so that every record can be identified.

A diagram that models the data of a system as entities, the attributes of each entity and the relationships between entities, including how many instances of each entity may take part in each relationship.

Everything outside the agent that it interacts with: it receives the agent's actions and responds with a new situation and a reward.

The effect of a technology on the natural environment over its life cycle, including the energy and water used to make and run it, the greenhouse gases emitted and the waste produced when it is discarded.

One complete pass of the whole training data set through a learning algorithm during training.

An action-selection rule in which an agent explores by choosing a random action with probability ε and otherwise exploits by choosing the action with the highest estimated value.

Fairness in how the benefits, costs and risks of a technology are distributed, taking into account that different groups start from different positions and may need different support to benefit equally.

Invalid test data, of the wrong type or clearly outside the valid range, that a program should reject without crashing.

The difference between the set point and the measured value of the controlled quantity, which the controller uses to decide its next output.

The family of standards (IEEE 802.3) for cabled local area networks, which defines the cables, connectors, signals and frame format that devices use to exchange data.

The study of what is right and wrong in human conduct, applied in computing to the choices made when designing, building and using technology and to their effects on people and the environment.

The straight-line distance between two points, the square root of the sum of the squared differences between their coordinates.

A guarantee that if no new updates are made to a data item, all copies of it will become identical after some time, although reads in the meantime may return an older value.

An event during the running of a program that disrupts the normal flow of instructions, represented by an object that records the type of error and a message.

A technique in which code that might fail is placed in a block that is monitored for exceptions, and handler code runs if a matching exception occurs, so the program can recover instead of stopping.

A file of machine code produced by translation that the operating system can load and run directly on one platform, without the source code or the translator.

The degree to which the reasons for a model's individual outputs can be described in terms a person can understand, such as which inputs most influenced a decision.

In reinforcement learning, choosing the action with the highest estimated reward based on what the agent has learned so far.

In reinforcement learning, choosing an action that is not currently believed to be the best, in order to gather information about its reward.

A complexity in which the number of steps roughly doubles each time the input size increases by one.

The extent to which a product can be extended with new features or adapted to new requirements without major changes to its existing design and code.

The process that copies data from source systems (extract), cleans and converts it into one consistent format (transform), and adds it to the data warehouse (load).

Using a model to predict values outside the range of the data it was fitted to, as opposed to interpolation, which predicts within that range.

The harmonic mean of precision and recall, 2 × precision × recall ÷ (precision + recall), which is high only when both are high.

The use of computer vision, usually a deep learning model, to detect a face in an image and identify or verify a person by comparing the face with stored images.

The central table of a star schema, holding one row per business event, such as a sale, with numeric measures (quantity, revenue) and a foreign key to each dimension table.

A design pattern in which a method or class creates and returns objects, choosing the concrete class from its input, so the code that asks for an object depends only on a superclass or interface and never names the concrete classes.

A design feature that makes a system move to, or stay in, a safe state when a component fails or power is lost.

Automatically switching a service to a standby server, link or device when the active one fails.

An example that is actually positive but that the model wrongly predicts as negative; a missed case.

An example that is actually negative but that the model wrongly predicts as positive; a false alarm.

The ability of a system to keep working correctly, possibly at reduced performance, when one or more of its components fail.

One measurable property of an example that is given to a machine learning model as an input, such as the number of links in an email or the floor area of a house.

Creating new features by combining or transforming existing ones, so that fewer or more informative features describe each record.

A score for each feature showing how much it contributes to a trained model's predictions.

The grid of values a convolutional layer produces by applying one kernel across its input, showing how strongly the kernel's pattern is present at each position.

Transforming the values of numerical features so that they lie on comparable scales, so that no feature dominates a model because of its units or range.

The process of choosing a subset of the existing features of a dataset that are most useful for predicting the target, and discarding the others.

Information about the output of a system, measured by sensors and returned to the controller as an input so that the output can be corrected.

A cycle in which a model's outputs influence the data later used to retrain or evaluate it, so that its errors or biases are reinforced over time.

The repeated sequence of steps by which the CPU fetches an instruction from main memory, decodes it in the control unit, and executes it, before moving on to the next instruction.

A cable that carries data as pulses of light along thin strands of glass or plastic, guided by total internal reflection.

One column of a table, holding a single property of the entity, with the same data type in every record.

A chip containing a large grid of logic blocks whose connections can be reconfigured after manufacture, so it can be programmed to act as a custom hardware circuit for a particular task and reprogrammed later.

A named collection of data held in secondary storage, which the operating system stores, finds and protects as a single unit.

A reference, returned when a file is opened, that a program uses to read from, write to and close that file.

The setting chosen when a file is opened that fixes which operations are allowed and how the existing contents are treated: read, write (replacing the contents) or append (adding to the end).

The location of a file in a file system, written as the sequence of directories leading to it followed by the file name, either from the root of the file system (absolute) or from the current working directory (relative).

The position in an open file at which the next read or write operation will start; it moves forward automatically after every read or write.

A server that stores files centrally and lets authorised users on the network open, save and share them.

The part of the operating system that organises data on secondary storage into named files and directories, records where each file's data is stored, and controls how files are created, read, changed and deleted.

The narrowing of the information a person sees online when personalisation algorithms keep showing content that matches their past behaviour and views, reducing exposure to other perspectives.

A feature selection strategy that ranks features by a statistical measure of their relationship with the target, such as correlation, calculated without training any model, and keeps the highest-ranked features.

A block attached to a try statement whose code always runs when the try statement ends, whether or not an exception occurred, typically to release resources such as open files.

Further training of a pre-trained model on a smaller, task-specific dataset, usually with a new output layer and small changes to the existing weights.

A hardware device or software program that monitors the network traffic entering and leaving a network or device and allows or blocks it according to a set of security rules.

A condition on fields of network traffic, such as direction, source or destination IP address, protocol and port, paired with an action that allows or denies the traffic that matches it.

Software stored in non-volatile memory on a hardware device that controls the device at a low level.

The rule that the item added earliest is the first to be removed.

The state of a table in which every field holds a single atomic value, there are no repeating groups, and every record is uniquely identified by a primary key.

A non-pre-emptive scheduling algorithm that runs processes in the order in which they arrive in the ready queue, each until its CPU burst is complete.

A function that gives each candidate solution in a genetic algorithm a score showing how well it solves the problem.

A text data type, written CHAR(n), that stores exactly n characters in every record, padding shorter values with spaces.

A numeric data type, written DECIMAL(p, s), that stores numbers exactly using p digits in total, s of them after the decimal point.

A Boolean variable that records whether an event has happened, set to an initial value before a loop and changed when the event occurs, so later code can test it.

Non-volatile memory with no moving parts that stores each bit as electric charge trapped in a memory cell, and that can be electrically erased and rewritten in blocks.

A database that stores all its data in a single table or file, with no links to other tables.

A design in which records in the same collection can have different fields and structures, so new attributes can be added without changing a table definition.

A data type for numbers that can have a fractional part, stored to a limited number of significant digits so that most values are approximations.

Floating-point operations per second, a measure of a processor's arithmetic speed; large values use prefixes such as tera ($10^{12}$), peta ($10^{15}$) and exa ($10^{18}$).

A diagram that represents an algorithm using standard symbols for each kind of step, joined by arrows that show the order in which the steps are carried out.

An arrow in a flowchart that joins two symbols and shows which step is carried out next.

To write all data waiting in an output buffer to the file on secondary storage, emptying the buffer.

A field in one table whose values must match values of the primary key in another table (or the same table), creating a link between the records.

A shape or pattern that contains smaller copies of itself at every scale, so that each part resembles the whole.

The unit of data at the network interface layer, made of a header with source and destination MAC addresses, the packet being carried, and a trailer with an error-checking value.

The number of frames (still images) shown per second of video, measured in frames per second (fps).

The end of a queue from which items are removed, and the operation that returns the item there without removing it; in an array-based queue, a pointer holds its index.

A layer of a neural network in which every neuron receives an input from every neuron in the previous layer, each connection having its own weight.

A relationship between fields in which each value of one field, or set of fields (the determinant), is always associated with exactly one value of another field, written determinant → dependent field.

A requirement that states a function or behaviour the system must provide, describing what the system does in response to particular inputs or events.

A horizontal timeline divided into labelled blocks that shows which task runs during each interval of time; in CPU scheduling, each block is a period in which one process uses the CPU.

The principle that the output of a computer system can be no better than its input, so a model trained on poor-quality data produces poor-quality predictions.

A network node that connects two networks and passes traffic between them, translating between different protocols or formats where the networks use different ones.

The European Union law, applied since 25 May 2018, that sets rules for processing personal data, gives individuals rights over their data, and allows fines of up to €20 million or 4% of a company's worldwide annual turnover, whichever is higher.

The ability of a trained model to make accurate predictions on new data that it did not see during training.

One complete cycle of a genetic algorithm (evaluation, selection, crossover and mutation), or the population that results from it.

Machine learning models that produce new content, such as text, images, audio or code, resembling the data they were trained on.

A search and optimisation method, inspired by natural selection, that improves a population of candidate solutions over many generations through selection of the fittest, crossover and mutation.

A system that captures, stores, analyses and displays data linked to locations on the Earth's surface, usually as layers on a map.

A measure of how mixed the classes in a node are, equal to the probability that two examples drawn at random from the node (with replacement) belong to different classes; 0 for a pure node.

The process by which a distributed database splits a query into sub-queries for the sites that hold the relevant data, runs them, and combines the partial results into one answer, choosing a plan that keeps network transfer low.

A variable declared outside all subprograms, which can be used anywhere in the program and exists for as long as the program runs.

An optimisation method that repeatedly changes a model's parameters by a small step in the direction that reduces the loss most steeply.

A NoSQL database that stores data as nodes (entities) connected by edges (relationships), both of which can have properties, and answers queries by following the edges.

A user interface in which the user interacts with the computer through visual elements such as windows, icons, menus and a pointer, or by touch.

A processor with a very large number of simple cores designed to carry out the same operation on many data items in parallel, originally for rendering graphics and now also for other highly parallel computations.

An ordering of binary values in which each value differs from the next in exactly one bit.

The clause that collects the rows of a result with equal values in the named columns into groups, so that the query returns one row for each group.

A logic circuit that adds two single bits, producing a sum bit (A XOR B) and a carry bit (A AND B).

A secondary storage device that records data magnetically on rapidly spinning disks (platters), using read/write heads on a moving arm.

A relationship in which an object of one class contains or holds a reference to one or more objects of another class as part of its state.

The situation in which two different keys produce the same index in a hash table.

A function that converts a key into a number used to choose its position in a hash table, always producing the same number for the same key.

A data structure that stores keys (or key–value pairs) in an array of buckets, using a hash function to calculate the bucket for each key so that it can be stored and found without searching.

The number produced by applying a hash function to a key; reduced modulo the table size, it gives the index of the key's bucket.

The clause that filters groups made by GROUP BY, keeping only the groups for which its condition is true.

The reference to the first node of a linked list, from which every traversal starts; it is null or None when the list is empty.

The number of edges on the longest path from the root node down to a leaf; a tree with only a root has height 0.

The base-16 number system, which uses sixteen digits, 0 to 9 and A to F (A = 10, B = 11, C = 12, D = 13, E = 14, F = 15), and column values that are powers of 16.

A layer of neurons between the input and output layers of a neural network, which transforms the outputs of the previous layer into new features.

Data organised as a tree, in which each item has one parent (except the item at the top) and may have any number of children, to any depth.

Memory designed to transfer very large amounts of data per second to a GPU or other processor, using a very wide connection and stacked memory chips placed beside the processor.

A programming language whose statements are written in terms of the problem being solved rather than the instructions of a particular processor, so that one statement can stand for many machine instructions.

The use of supercomputers, very large clusters of processors connected by fast networks and run as one system, to solve problems too large for ordinary computers.

One step in a packet's journey, from one router (or host) to the next device along the route.

Increasing a system's capacity by adding more servers (nodes) and spreading the data and the workload across them.

A part of a program, such as a method or a loop, that runs so often that it accounts for a large share of the program's execution time.

HTTP sent through an encrypted connection created by Transport Layer Security (TLS), so that requests and responses cannot be read or altered in transit and the server's identity is checked.

A three-digit number in an HTTP response that reports the result of the request, with the first digit giving its class (2 for success, 3 for redirection, 4 for a client error, 5 for a server error).

A central connecting device that repeats every signal it receives out of all its other ports, without reading any addresses.

A design in which a person reviews, approves or can override an automated system's output before it takes effect, especially for decisions with serious consequences.

A topology that combines two or more basic topologies, such as star networks joined to each other through a mesh or star backbone.

A setting chosen before training that controls how a model learns or how complex it is, and that is not learned from the training data.

The process of searching for the hyperparameter values that give a model the best performance on validation data.

The application-layer protocol of the World Wide Web, in which a client sends a request for a resource and a server returns a response containing a status code and, usually, the resource.

Software that creates and runs virtual machines, sharing the physical computer's CPU, memory, storage and network between them and keeping them isolated from each other.

The rule that ANDing or ORing a variable with itself gives the same variable: $A \cdot A = A$ and $A + A = A$.

The name a programmer gives to a variable, constant, subprogram or class so that it can be referred to in code.

The rule that ANDing a variable with 1, or ORing it with 0, leaves it unchanged: $A \cdot 1 = A$ and $A + 0 = A$.

The number of pixels in an image, usually given as width × height.

The property of a value that cannot be changed after it is created, so any change produces a new value instead.

A programming language in which the programmer writes the sequence of steps, such as assignments, loops and selections, that the computer carries out in order.

Values that are recorded in inconsistent or unusable forms, such as mixed date formats, mixed units, inconsistent spelling or capitals, or numbers stored as text.

Handling missing data by replacing each missing value with an estimate, such as the mean, median or mode of the feature, or the previous value in a time series.

A database that keeps its data in main memory (RAM) rather than on disk, giving very low latency for reads and writes.

The node that comes immediately after a given node in an in-order traversal; for a node with a right subtree, it is the smallest value in that right subtree.

A traversal of a binary tree that visits the left subtree, then the node itself, then the right subtree, applied recursively at every node.

An algorithm that transforms its data within the memory the data already occupies, using only a constant amount of extra memory, $O(1)$.

Values that are present but do not match reality, such as impossible values, values that break a rule of their field, or values that contradict other fields in the same record.

The input variable in a model, whose values are used to predict another variable; also called the predictor.

A number that identifies the position of an item in a string or other sequence, counting from 0 for the first item.

The process of recreating an index from the current data, or reorganising its pages, so that it is compact and in order again after many changes to the table.

The freedoms and protections each person is entitled to, such as privacy, freedom of expression, equal treatment and due process, including how these apply to digital technology and data.

The use of a trained model to produce an output, such as a prediction or classification, for new input data that the model has not seen before.

A loop whose terminating condition is never met, so it repeats without end until the program is stopped from outside.

The reduction in entropy (a measure of how mixed the classes are) achieved by splitting a node on a feature; the split with the highest information gain is chosen.

The principle that a class keeps its data and implementation details inaccessible to other code, which can use the object only through the methods the class chooses to make public.

A cloud service model in which the provider supplies virtualised computing resources (virtual machines, storage and networking), and the customer installs and manages the operating system, runtime, applications and data.

The mechanism by which a new class is defined from an existing class, automatically gaining its attributes and methods and adding or changing some of its own.

The first layer of a neural network, which receives the input features, one neuron per feature, and passes them to the next layer.

A description of the data a solution receives: what each item is, where it comes from, its data type, its format and its range of valid values.

The parallelogram-shaped flowchart symbol for data entering the algorithm (input) or results leaving it (output).

A network protocol that sends data, including login details, without encryption or without checking the identity of the other party, so that it can be read or altered in transit.

A fault in a poorly designed table where a fact cannot be recorded without also recording an unrelated fact, because part of the primary key would otherwise be empty.

An object considered as a member of the class it was created from.

A method that is called on a particular object and can read and change that object's instance variables through its self reference.

An attribute of which each object holds its own separate copy, so that its value can differ from object to object.

The process of creating a new object from a class, which allocates memory for the object and runs its constructor.

A register that holds the instruction currently being decoded and executed.

A data type for whole numbers, positive, negative or zero, with no fractional part.

Division that gives only the whole-number part of the quotient and discards the remainder.

Data combined from several sources and converted into consistent formats, names, codes and units, so that the same fact is represented the same way whatever its source.

A software application that brings together the tools for writing, running and debugging programs, such as a code editor, a translator and a debugger.

Testing components of a product together, through their real connections, to check that data passes correctly between them and the combined behaviour is right.

A rule stored in the database schema that the DBMS checks on every insert, update and delete, rejecting any change that would break it.

In a regression equation, the predicted value of the dependent variable when the independent variable is zero.

A named set of method declarations, with no instance variables, that a class promises to implement in full, so objects of unrelated classes can be used through the same methods.

The coursework component of IB Computer Science, in which a student specifies a problem of their own choosing, then plans, designs, codes, tests and evaluates a software product that solves it, documenting each stage against five criteria.

The global system of interconnected computer networks that exchange data using the TCP/IP family of protocols.

The TCP/IP layer that gives each host a logical IP address and moves packets from the source host to the destination host across any number of networks, with routers choosing the path.

Everyday physical objects fitted with processors, sensors and network connections so that they can collect and exchange data without a person operating them.

The protocol at the internet layer of the TCP/IP model that gives each host a logical address and carries data between hosts, across any number of networks, as independently addressed packets.

An organisation that connects its customers' devices and networks to the internet, usually for a fee.

The degree to which a person can understand how a model reaches its predictions.

A translator that analyses and executes a program one statement (or instruction) at a time while it runs, without producing a separate machine-code file.

The difference between the upper quartile and the lower quartile of a set of values, $Q_3 - Q_1$, which measures the spread of the middle 50% of the data.

A signal from a device or program to the CPU indicating that an event needs immediate attention, causing the CPU to suspend its current task and run code that handles the event.

A short routine, usually part of the operating system or a device driver, that the CPU runs in response to a particular interrupt in order to deal with the event that caused it.

The set of elements that are in both of two sets; written $A \cap B$.

A system that monitors network traffic or device activity for signs of attack and alerts administrators, without blocking the traffic itself.

A system placed in the path of network traffic that detects signs of attack and blocks or drops the malicious traffic automatically.

A logical numeric address, assigned by software, that identifies a network interface on an IP network and the network it belongs to.

Version 4 of the Internet Protocol, which uses 32-bit addresses written in dotted decimal notation as four numbers from 0 to 255.

The depletion of the pool of unallocated IPv4 addresses, so that new devices and networks can no longer be given new public IPv4 addresses freely.

Version 6 of the Internet Protocol, which uses 128-bit addresses written as eight groups of four hexadecimal digits separated by colons.

Records outside the population a model is meant for, or fields that carry no useful information about what the model must predict.