Key terms bank

Abstraction — Hiding detail so only the essential features are shown.

Acceptance testing — Client/users check the system meets their needs.

Accountability — Actions are traceable to a person (logs).

Actuator — Output device that moves or acts (motor, hydraulic actuator).

Adaptive maintenance — Changing software for a new environment.

Agile — Iterative approach with short sprints and continuous feedback.

AI — Machines performing tasks that need human intelligence; ML is a subset.

API — Rules letting software components request data or services from each other.

Asymmetric encryption — Public key encrypts, private key decrypts.

Authentication — Proving who you are.

Authorisation — Deciding what you may do.

Availability — Systems and data are accessible when needed.

Backtracking — Try an option, retreat on a dead end, try another.

Beta testing — Real users trial a near-final version.

Bias (human, dataset) — Unfair skew in data, labels or decisions that favours some outcomes or groups.

Big data — Huge, fast, varied data sets.

Boundary testing — Test at and either side of the limits.

BPA — Business process automation of whole workflows.

Breakpoint — Pauses execution at a chosen line.

Business continuity — Keeping critical operations running through disruption.

Cipher text — Encrypted, unreadable data. Plain text: readable data.

Class — Blueprint for objects.

CMS — Content management system (WordPress, Drupal, Joomla).

Code review — People read code to find flaws.

Confidentiality — Only authorised people can read data.

Corrective maintenance — Fixing bugs after release.

CRUD — Create, Read, Update, Delete (INSERT, SELECT, UPDATE, DELETE).

CSRF — Cross-site request forgery: forged request using a logged-in user's trust.

CSS — Cascading Style Sheets: presentation of web pages.

DAST — Dynamic application security testing of the running app.

Data dictionary — Table describing every variable (type, format, size, description, example, validation).

Data mining — Finding patterns in large data sets.

Data packet — Chunk of data with header and payload.

Decision tree — Diagram of decisions and resulting actions (also an ML model).

DevOps — Development plus operations: CI/CD, rapid feedback, automation, shared ownership.

DFD — Data flow diagram.

Digital signature — Hash encrypted with sender's private key; proves origin and integrity.

Direct implementation — Old system off, new on at once.

Disaster recovery — Restoring systems and data after a major incident (RTO, RPO).

DNS — Domain name system: name to IP address (port 53).

Driver — Test code that calls a module under test.

Encapsulation — Data and methods bundled and protected.

Exception management — Handling runtime errors without crashing (try/except).

Feasibility — Whether a project is achievable (schedule, financial).

FTP / SFTP — File transfer (port 21) / secure file transfer over SSH (port 22).

Functional requirement — What the system must do.

Gantt chart — Tasks against time with dependencies and milestones.

Generalisation — Common features pulled up into a superclass.

GDPR — EU General Data Protection Regulation.

Hardening — Reducing attack surface through configuration and patching.

Hash value — Fixed-length output of a one-way function.

HTTP / HTTPS — Web protocol (port 80) / encrypted version (port 443).

IMAP / POP3 / SMTP — Email: sync on server (143) / download (110) / send (25).

Inheritance — Subclass reuses and extends a superclass.

Input validation — Checking input meets rules before use.

Integrity — Data accurate and unchanged by unauthorised parties.

IPv4 — 32-bit address written as four numbers 0-255.

K-nearest neighbour — Classify by the majority of the K closest points.

Language-dependent optimisation — Efficiency gains using features of a specific language.

Lazy loading — Load media only when needed.

Linear regression — Predicts a continuous numeric value on a straight line.

Logbook / process diary — Dated record of progress, issues and reflections.

Logistic regression — Classification (yes/no) using probability.

Machine learning — Algorithms that learn patterns from data.

Metadata — Data about data.

Minification — Removing unnecessary characters from code to reduce file size.

MLOps — Design, model development and operations stages for ML.

Neural network — Layers of weighted nodes trained by backpropagation.

ORM — Object-relational mapping: database rows as objects.

Parallel implementation — Old and new run side by side.

Penetration testing — Authorised simulated attack that exploits weaknesses.

Perfective maintenance — Improving performance or features.

Phased implementation — New system introduced in stages.

Pilot implementation — New system trialled by a small group first.

Polymorphism — Same method name, different behaviour per class.

Privacy by design — Proactive not reactive; embed privacy into design; respect for user privacy.

Progressive web app (PWA) — Installable, offline-capable web app (service worker, manifest, HTTPS).

Race condition — Outcome depends on the timing of concurrent access to shared data.

Regulatory compliance — Meeting laws, regulations and standards.

Reinforcement learning — Learn from rewards and penalties.

RPA — Robotic process automation: bots mimic user actions.

Sandboxing — Isolated environment that limits what code can do.

SAST — Static application security testing of source code without running it.

Sanitisation — Cleaning or encoding input to neutralise dangerous characters.

Security by design — Security built in from the start.

Semi-supervised learning — Few labels plus many unlabelled examples.

Sensor — Input device detecting the environment.

Service worker — Background script that caches and intercepts requests.

Session management — Tracking a user's state across requests with a session ID.

Side-channel attack — Learning secrets from timing, power or electromagnetic leakage.

SQL injection — Malicious SQL inserted via input; stopped by parameterised queries.

SSL/TLS certificate — Proves a site's identity and enables HTTPS.

Stub — Placeholder for an unfinished module.

Supervised learning — Learn from labelled data.

Symmetric encryption — One shared key.

TCP/IP — Reliable delivery (TCP) and addressing/routing (IP).

Two's complement — Signed binary: invert bits and add 1.

Unsupervised learning — Find patterns in unlabelled data.

Version control — Tracks changes, branches, merges and reverts code.

Volume testing — Test with large amounts of data.

W3C / WAI — Web standards body / web accessibility initiative.

Waterfall — Sequential, plan-driven phases.

WAgile — Hybrid of Waterfall planning and Agile iteration.

Web manifest — JSON file with app name, icons and display mode.

XSS — Cross-site scripting: malicious script injected into a page other users view.