Web, Mobile and Interface Development
user-centered design
noun
Meaning
An approach that puts users' needs, goals and limits at the centre of every stage of design, using research and testing with real people to guide decisions.
Where you use it
Shortened to UCD. Appears in job ads, process documents and case studies. The international standard ISO 9241-210 uses the closely related term 'human-centred design'. People say a decision is, or isn't, user-centered.
Examples
Our process follows user-centered design: we research, design, test and repeat.
Adding a feature just because the CEO likes it isn't exactly user-centered design.
Usage note
British English writes user-centred and human-centred. Both spellings are correct; just be consistent.
Common phrases
user-centered design process · user-centered approach · principles of user-centered design
More about this term →pain point
noun
Meaning
A specific problem, frustration or obstacle that users experience, such as a confusing step, a long wait or a missing feature.
Where you use it
Heard in research readouts, journey maps and product meetings: 'the biggest pain point is delivery tracking'. Teams identify, uncover and address pain points. The term is common in sales and marketing too.
Examples
The main pain point in our interviews was the long sign-up form.
This feature addresses a real pain point for small shop owners.
Common phrases
identify pain points · address a pain point · key pain point · customer pain points
More about this term →button
noun
Meaning
A clickable or tappable element that lets users perform an action, such as submitting a form, saving changes or opening a menu.
Where you use it
Design systems define button types such as primary, secondary, tertiary (text) and destructive buttons, plus states like hover, pressed and disabled. Labels should describe the action: 'Save changes' rather than 'OK'.
Examples
Make 'Pay now' the primary button and 'Back' a secondary one.
Users tapped the disabled button again and again because it didn't look disabled.
Usage note
Pronounced /ˈbʌtn/, roughly 'BAT-n', with no clear vowel in the second syllable, not like Turkish 'buton'.
Common phrases
primary button · disabled button · button label · tap the button
More about this term →call to action
noun
Meaning
A button, link or short message that asks users to take the next step, such as 'Sign up', 'Add to cart' or 'Start free trial'.
Where you use it
Almost always shortened to CTA, said letter by letter. Designers and marketers talk about the primary CTA, CTA placement and CTA copy. A screen usually works best with one clear main call to action.
Examples
Four buttons are competing for attention; this page needs one clear call to action.
Changing the call to action from 'Submit' to 'Get my quote' increased clicks.
Usage note
The plural is calls to action, not call to actions.
Common phrases
primary call to action · clear call to action · CTA button · CTA copy
More about this term →navigation
noun
Meaning
The menus, links, tabs and other elements that let people move around a website or app and understand where they are.
Where you use it
Designers talk about global navigation, the navigation bar (navbar), bottom navigation on mobile and side navigation. Often shortened to 'nav' in casual talk and in file names. The verb is navigate: 'users navigate to Settings'.
Examples
Users couldn't find their orders because 'My account' was hidden in the navigation.
On mobile we moved the main navigation to a bottom tab bar.
Common phrases
main navigation · navigation bar · bottom navigation · navigation menu
More about this term →handoff
noun
Meaning
The stage when designers pass finished designs, specs and assets to developers so they can build them, ideally with a walkthrough and clear notes.
Where you use it
Tools such as Figma's Dev Mode show measurements, colours and code snippets for handoff. Teams talk about a smooth handoff, handoff notes and handoff meetings. The verb is two words: 'hand off the designs'.
Examples
The handoff is on Monday, so every screen needs final specs by Friday.
A good handoff includes empty states and error messages, not just the happy path.
Usage note
Also written hand-off. The noun is one word (handoff); the verb is two (hand off).
Common phrases
design handoff · developer handoff · smooth handoff · handoff notes
More about this term →user interview
noun
Meaning
A one-to-one conversation with a user or potential user, usually 30 to 60 minutes long, to learn about their goals, habits, problems and opinions.
Where you use it
Researchers conduct user interviews with a discussion guide of open questions ('Tell me about the last time you…'). Good interviewers avoid leading questions and ask about past behaviour rather than future wishes.
Examples
I've scheduled six user interviews with restaurant owners next week.
In the user interviews, nobody mentioned price; they all talked about delivery time.
Usage note
A leading question suggests the answer ('Don't you think this is easier?'). Ask neutral questions instead ('How did that feel?').
Common phrases
conduct user interviews · interview guide · in-depth interview · interview notes
More about this term →qualitative research
noun
Meaning
Research that collects words, observations and stories from a small number of people to understand why they think and behave the way they do.
Where you use it
Includes interviews, usability tests, diary studies and field visits. Research teams often shorten it to 'qual' ('we need some qual on this'). It answers why and how, not how many.
Examples
The numbers show a drop at checkout, so we'll do qualitative research to understand why.
Eight interviews are enough for qualitative research at this stage.
Usage note
Qualitative and quantitative are easy to mix up. Remember that quantitative contains 'quantity', meaning numbers.
Common phrases
qualitative research methods · qualitative data · qualitative insights · qual and quant
More about this term →quantitative research
noun
Meaning
Research that collects numbers from many users, such as survey ratings, click rates or task times, to measure behaviour and spot patterns.
Where you use it
Includes surveys, analytics, A/B tests and benchmark studies. Shortened to 'quant'. Teams usually combine it with qualitative research: quant shows what is happening, qual explains why.
Examples
Our quantitative research shows that 60% of users never open the settings page.
Quantitative research tells us how many users drop off, not why they leave.
Common phrases
quantitative research methods · quantitative data · statistically significant · qual and quant
More about this term →card sorting
noun
Meaning
A research method in which participants group topics written on cards into categories that make sense to them, showing how users expect content to be organised.
Where you use it
In an open card sort, participants create and name their own groups; in a closed card sort, the categories are given. It can be done with paper cards or online tools and is used when designing menus and categories.
Examples
Card sorting showed that users expect 'Returns' under 'Orders', not under 'Help'.
We did online card sorting with 30 participants before redesigning the menu.
Common phrases
open card sorting · closed card sort · run a card sort · card sorting session
More about this term →low-fidelity
adjective
Meaning
Describes an early design, wireframe or prototype that is rough and simple, without real colours, images or detailed content, so it is quick to make and change.
Where you use it
Often shortened to lo-fi ('a lo-fi prototype'). Paper sketches and grey-box wireframes are low-fidelity. Teams use them to explore many ideas cheaply and to keep feedback on structure, not colours.
Examples
Start with low-fidelity sketches; it's too early to choose colours.
We tested a low-fidelity prototype made of paper screens.
Usage note
Fidelity here means how close something is to the final product, not loyalty. Stress the second syllable: fi-DEL-i-ty.
Common phrases
low-fidelity wireframe · low-fidelity prototype · lo-fi sketch · keep it low-fidelity
More about this term →high-fidelity
adjective
Meaning
Describes a design or prototype that looks and behaves almost like the final product, with real content, colours, images and interactions.
Where you use it
Shortened to hi-fi. High-fidelity mockups and prototypes are used for final reviews, realistic usability tests and developer handoff. They take longer to make, so teams usually create them after the structure is agreed.
Examples
The high-fidelity prototype was so realistic that some participants thought the app was already live.
Don't jump to high-fidelity screens before the flow is approved.
Common phrases
high-fidelity mockup · high-fidelity prototype · hi-fi designs · move to high fidelity
More about this term →clickable prototype
noun
Meaning
A prototype in which screens are linked together, so clicking or tapping buttons and links takes the user to the next screen as in a real product.
Where you use it
Made by connecting frames in tools like Figma and shared as a link for usability tests or stakeholder reviews. Also called an interactive prototype. Usually only the main paths are clickable, so testers need clear tasks.
Examples
I'll send you a clickable prototype so you can try the booking flow on your phone.
Only the checkout screens are clickable in this prototype.
Common phrases
share a clickable prototype · clickable prototype link · interactive prototype · make the screens clickable
More about this term →feedback
noun
Meaning
The signal a system gives in response to a user's action, such as a button changing colour, a loading spinner or a success message, so people know what happened.
Where you use it
A basic principle of interaction design: every action should get clear, quick feedback. The same word also means comments on someone's work: designers give, get and ask for feedback in reviews.
Examples
There's no feedback after tapping 'Upload', so people tap it three times.
Can you give me some feedback on these two layouts before the review?
Usage note
Feedback is uncountable: say 'some feedback' or 'a piece of feedback', never 'feedbacks'.
Common phrases
visual feedback · instant feedback · give feedback · feedback message
More about this term →design critique
noun
Meaning
A meeting where designers show work in progress and colleagues give structured, honest feedback on whether the design meets its goals.
Where you use it
Often shortened to 'crit'. Good critique focuses on goals and users ('Does this help a first-time buyer?') rather than personal taste ('I don't like blue'). Many teams hold a weekly design critique.
Examples
I'll bring the new dashboard to Thursday's design critique.
In a design critique, explain the problem first, then show the solution.
Usage note
Critique (pronounced kri-TEEK) means careful evaluation, not an attack. Criticism sounds more negative.
Common phrases
weekly design critique · run a critique · give critique · critique session
More about this term →affordance
noun
Meaning
A quality of an object or interface element that suggests how it can be used, for example a raised button that looks pressable or a handle that invites pulling.
Where you use it
The idea comes from psychologist James Gibson and was brought into design by Don Norman. Designers say a flat link 'has poor affordance' when it doesn't look clickable. Norman later added 'signifier' for the visible clue itself.
Examples
The card has no affordance, so nobody realised it could be tapped.
Underlined blue text is a strong affordance for a link.
Common phrases
clear affordance · poor affordance · visual affordance · affordances and signifiers
More about this term →style guide
noun
Meaning
A document that sets the visual and writing rules for a brand or product, such as colours, typography, logo use, icon style and tone of voice.
Where you use it
Designers follow or stick to the style guide so that screens stay consistent. In product teams it is often part of a larger design system; content teams may keep a separate writing style guide for spelling and tone.
Examples
The style guide says we only use orange for primary actions.
Before designing the campaign page, read the brand style guide.
Common phrases
brand style guide · follow the style guide · UI style guide · writing style guide
More about this term →color contrast
noun
Meaning
The difference in brightness between text or an icon and its background, which decides how easy it is to see, especially for people with low vision.
Where you use it
Measured as a contrast ratio. WCAG level AA asks for at least 4.5:1 for normal text and 3:1 for large text; level AAA asks for 7:1. Designers test colour pairs with contrast checker plugins.
Examples
Light grey text on white fails the colour contrast check.
We darkened the button so the color contrast meets WCAG AA.
Usage note
Colour is the British spelling and color the American one. Both appear in design work, but tools and code use color.
Common phrases
low color contrast · check the contrast ratio · meet contrast requirements · high-contrast mode
More about this term →WCAG
abbreviation
Meaning
Short for Web Content Accessibility Guidelines: the international standard, published by the W3C, that explains how to make websites and apps accessible.
Where you use it
Built on four principles, remembered as POUR: perceivable, operable, understandable and robust. Success criteria have three levels, A, AA and AAA; most laws and companies aim for AA. Version 2.2 was published in 2023.
Examples
Our contract requires the website to meet WCAG 2.2 level AA.
This error message fails WCAG because it relies only on red colour.
Usage note
Usually pronounced as one word, 'wuh-cag', though some people say the letters. W3C stands for World Wide Web Consortium.
Common phrases
meet WCAG · WCAG AA compliance · WCAG success criteria · WCAG guidelines
More about this term →white space
noun
Meaning
The empty space between and around elements on a page, which gives content room to breathe, groups related items and makes the design easier to read.
Where you use it
Also called negative space. It doesn't have to be white; any empty area counts. Designers say a screen 'needs more white space' or 'feels cramped'. The spacing values in a design system control it.
Examples
The form feels cramped; add more white space between the sections.
Generous white space makes the pricing table much easier to scan.
Usage note
Also written as one word, whitespace. In programming, whitespace means spaces, tabs and line breaks in text, which is a different idea.
Common phrases
generous white space · use white space · add more white space · negative space
More about this term →grid
noun
Meaning
A set of invisible columns, rows and gaps that designers use to place and align elements consistently across a page or screen.
Where you use it
A common web setup is a 12-column grid on desktop and a 4-column grid on mobile. Designers talk about columns, gutters, margins and 'breaking the grid'. Many teams also use an 8-point spacing grid.
Examples
The image doesn't sit on the grid, so the whole page looks slightly off.
We use a 12-column grid with 24-pixel gutters.
Common phrases
12-column grid · layout grid · align to the grid · break the grid
More about this term →color palette
noun
Meaning
The fixed set of colours chosen for a product or brand, usually with primary, secondary and neutral colours plus status colours such as red for errors and green for success.
Where you use it
Defined in the style guide or design system, often with several shades of each colour (blue 100 to blue 900). Designers check every combination for contrast and test the palette in dark mode as well.
Examples
Please pick colours from our color palette instead of using random hex codes.
The new color palette has fewer brand colours and more neutral greys.
Usage note
British English writes colour palette. Don't confuse palette (colours) with pallet (a wooden platform for goods); they sound the same.
Common phrases
brand color palette · limited color palette · neutral palette · define a color palette
More about this term →icon
noun
Meaning
A small picture or symbol that stands for an action, object or idea, such as a magnifying glass for search or a bin for delete.
Where you use it
Designers use icon sets with one consistent style (outline or filled) and fixed sizes such as 16, 20 or 24 pixels. Icons without labels can be confusing, so important actions usually get a text label too.
Examples
Half of the participants didn't know what the star icon meant.
Add a label next to the icon; a picture alone isn't enough here.
Usage note
Pronounced EYE-kon: it starts with the sound of 'eye', not the 'i' of Turkish 'ikon'.
Common phrases
icon set · icon button · outline icon · icon with a label
More about this term →mental model
noun
Meaning
A person's own idea of how something works, based on past experience, which shapes what they expect a product to do.
Where you use it
Designers try to match the product to users' mental models, for example by using a shopping cart because people know it from real shops. Problems start when the system works differently from what users believe.
Examples
In users' mental model, 'Save' also sends the document, so they never press 'Share'.
The interviews helped us understand customers' mental model of how insurance works.
Common phrases
match users' mental models · mental model of · mismatched mental model · build a mental model
More about this term →cognitive load
noun
Meaning
The amount of mental effort a person needs to understand and use something; too much of it makes people slow, confused or likely to give up.
Where you use it
Designers reduce cognitive load by showing fewer options at once, using familiar patterns, grouping information and remembering data for the user. The idea comes from educational psychology.
Examples
Twelve plan options create too much cognitive load; let's show three.
Pre-filling the address reduces cognitive load at checkout.
Common phrases
reduce cognitive load · high cognitive load · minimise cognitive load · cognitive overload
More about this term →friction
noun
Meaning
Anything in an experience that slows users down or makes a task harder, such as extra steps, long forms, unclear labels or slow loading.
Where you use it
Teams try to remove or reduce friction in key flows like sign-up and checkout. Sometimes designers add friction on purpose, for example a confirmation step before deleting an account. The adjective is frictionless.
Examples
Asking for a phone number at sign-up adds friction we don't need.
We removed friction from checkout by allowing guest orders.
Common phrases
reduce friction · add friction · points of friction · frictionless experience
More about this term →empty state
noun
Meaning
What a screen shows when there is no content yet, such as an empty inbox, a search with no results or a new account with no projects.
Where you use it
A good empty state explains what will appear there and offers a clear next step ('Create your first project'), often with a small illustration. It is easy to forget during design, so reviewers often ask for it.
Examples
The empty state on the favourites page just says 'No items', which isn't very helpful.
Design the empty state too, because every new user sees it first.
Common phrases
design the empty state · empty state illustration · helpful empty state · zero-results state
More about this term →tooltip
noun
Meaning
A small text box that appears when users hover over, focus on or tap an element, giving a short explanation or the name of an icon.
Where you use it
Useful for labelling icon buttons or explaining a term. Touchscreens have no hover, so important information should not be hidden only in a tooltip. Tooltips should also open with keyboard focus.
Examples
Add a tooltip to the info icon that explains what 'net price' means.
On mobile, nobody sees the tooltip because there's no hover.
Common phrases
show a tooltip · tooltip text · hover tooltip · info tooltip
More about this term →checkbox
noun
Meaning
A small square that users click or tap to select or clear an option; several checkboxes in a group can be selected at the same time.
Where you use it
Used for independent choices and for agreeing to terms. When only one option can be chosen, designers use radio buttons instead. Checkboxes have checked, unchecked and sometimes indeterminate states.
Examples
Users can filter by several sizes, so use checkboxes, not radio buttons.
Make the whole label clickable, not just the tiny checkbox.
Usage note
British speakers often say 'tick the box'; Americans say 'check the box'. A pre-checked box for marketing emails can be seen as a dark pattern.
Common phrases
tick the checkbox · check a checkbox · checkbox list · pre-checked checkbox
More about this term →toggle
noun
Meaning
A switch that turns a setting on or off with one tap and usually takes effect immediately, like turning on dark mode or notifications.
Where you use it
Also called a toggle switch or just a switch. It suits settings that apply instantly; inside a form with a 'Save' button, a checkbox is usually clearer. Also a verb: 'toggle the setting on'.
Examples
Use a toggle for 'Email notifications', because the change saves immediately.
It isn't clear whether the toggle is on or off; the colours are too similar.
Common phrases
toggle switch · toggle on · toggle off · toggle between
More about this term →breadcrumb
noun
Meaning
A row of links, usually near the top of a page, that shows where the current page sits in the site structure, such as Home > Shoes > Running.
Where you use it
Usually plural: breadcrumbs or breadcrumb navigation. Common on e-commerce and documentation sites with deep hierarchies, less so in mobile apps. The name comes from the fairy tale where children drop breadcrumbs to find their way home.
Examples
Add breadcrumbs so shoppers can jump back to the category page.
The breadcrumb trail shows the full path, but it's too small to tap on mobile.
Usage note
The b in crumb is silent: breadcrumb is pronounced BRED-krum.
Common phrases
breadcrumb navigation · breadcrumb trail · add breadcrumbs · breadcrumb link
More about this term →design token
noun
Meaning
A named value that stores a design decision, such as a colour, font size or spacing amount, so designers and developers use the same value everywhere.
Where you use it
Instead of a raw value like #1A73E8, teams use a token like color-primary or spacing-md. Changing one token updates every screen that uses it, which makes theming and dark mode easier. In Figma, tokens are often stored as variables.
Examples
Don't hard-code the blue; use the color-primary design token.
We updated the spacing design tokens, and every screen changed at once.
Common phrases
color tokens · spacing tokens · token naming · update a design token
More about this term →UI kit
noun
Meaning
A ready-made collection of interface elements, such as buttons, forms, icons and navigation bars, that designers drag into their files to build screens quickly.
Where you use it
Some UI kits are free or paid community files; others are a company's own kit built from its design system. Apple and Google publish official UI kits for iOS and Android so designers can match native components.
Examples
I used the iOS UI kit so the settings screen feels native.
Our UI kit is out of date; half the buttons don't match the code.
Usage note
A UI kit is mainly a design file; a design system also includes code, rules and documentation.
Common phrases
free UI kit · iOS UI kit · update the UI kit · UI kit components
More about this term →spacing
noun
Meaning
The amount of space between and inside elements, such as the gap between cards or the padding inside a button, chosen to create rhythm and grouping.
Where you use it
Many teams use a spacing scale based on multiples of 4 or 8 pixels (4, 8, 16, 24, 32) and store it as tokens. Common partner words: padding (inside), margin (outside) and gap (between items).
Examples
The spacing between the cards is 16 pixels on one page and 20 on another.
Use the spacing tokens instead of typing random numbers.
Common phrases
spacing scale · consistent spacing · vertical spacing · 8-point spacing system
More about this term →pixel-perfect
adjective
Meaning
Matching the design exactly, down to every pixel of size, spacing and colour, especially when a built product is compared with the original design file.
Where you use it
Used in handoff and design QA: 'Is the build pixel-perfect?' Some teams find it unrealistic for responsive products with many screen sizes and aim for consistency with the design system instead.
Examples
The client expects a pixel-perfect match with the approved mockups.
It doesn't need to be pixel-perfect; it needs to use the right components.
Common phrases
pixel-perfect design · pixel-perfect implementation · pixel-perfect match · design QA
More about this term →microinteraction
noun
Meaning
A small, single-purpose moment of interaction with its own feedback, such as liking a post, switching a toggle or a field shaking after a wrong password.
Where you use it
The term was made popular by Dan Saffer's book 'Microinteractions'. Designers use small animations and haptics to make these moments clear and pleasant, but too many moving details can feel distracting or slow.
Examples
The heart animation when you like a photo is a classic microinteraction.
Let's add a microinteraction so users see that the item went into the cart.
Usage note
Also written micro-interaction, with a hyphen.
Common phrases
subtle microinteraction · design microinteractions · microinteraction animation · delightful details
More about this term →hover state
noun
Meaning
The way an element looks when the mouse pointer is over it, for example a button getting darker, which shows that it can be clicked.
Where you use it
One of the standard component states, along with default, focus, pressed (active) and disabled. Designers define every state in the design system. Touchscreens have no hover, so it should never carry essential information.
Examples
The links have no hover state, so they don't feel clickable on desktop.
Please add hover states for the table rows before handoff.
Common phrases
add a hover state · hover effect · on hover · button states
More about this term →gesture
noun
Meaning
A movement of the fingers on a touchscreen, such as a tap, swipe, pinch or long press, that tells a device to do something.
Where you use it
Mobile designers talk about tap, double-tap, swipe, drag, pinch to zoom and long press. Gestures are invisible, so important actions should also have a visible button. iOS and Android differ in some system gestures.
Examples
Users didn't discover the swipe gesture for deleting messages.
Our custom gesture conflicts with the Android back gesture.
Usage note
Pronounced JES-cher. In everyday English, gesture also means a hand movement or a kind act ('a nice gesture').
Common phrases
swipe gesture · gesture navigation · pinch gesture · hidden gesture
More about this term →microcopy
noun
Meaning
The short pieces of text in an interface, such as button labels, hints, error messages and confirmations, that guide users through a task.
Where you use it
Written by UX writers or content designers, often together with product designers. Small changes in microcopy can have a big effect on conversion and on the number of support tickets. The wider practice is called UX writing.
Examples
The microcopy under the password field explains the rules before users make a mistake.
Can you review the microcopy on the payment screen? It sounds too formal.
Usage note
Microcopy is uncountable: say 'some microcopy', not 'a microcopy'.
Common phrases
write microcopy · UX microcopy · button microcopy · helpful microcopy
More about this term →card
noun
Meaning
A rectangular container that groups related content about one item, such as a product photo, title and price, and often links to more details.
Where you use it
Cards are common in feeds, dashboards and product lists because they fit well into responsive grids. Designers decide whether the whole card is clickable or only a button inside it, and keep card heights consistent.
Examples
Each recipe card shows a photo, the cooking time and a save button.
On mobile, the cards stack in a single column.
Common phrases
product card · card layout · clickable card · card component
More about this term →focus state
noun
Meaning
The visible style, usually an outline or ring, that shows which element is currently selected when someone moves through a page with a keyboard.
Where you use it
Keyboard and switch users depend on it to know where they are. Removing the browser's focus outline without a replacement is a common accessibility failure, and WCAG requires a visible focus indicator.
Examples
The focus state is almost invisible on the dark header.
Every interactive component in the design system needs a clear focus state.
Common phrases
visible focus state · focus ring · focus indicator · design the focus state
More about this term →inclusive design
noun
Meaning
An approach that designs for the full range of human diversity, including ability, age, language, culture and situation, so that more people can use a product.
Where you use it
Broader than accessibility: it also considers temporary and situational limits, such as a broken arm or bright sunlight. Microsoft's inclusive design toolkit helped make the idea widely known. Teams include diverse participants in research.
Examples
Inclusive design means testing with older adults, not just colleagues in their twenties.
Captions are a good example of inclusive design: they help deaf users and people watching in noisy places.
Common phrases
inclusive design principles · practise inclusive design · inclusive by default · design for diversity
More about this term →tree testing
noun
Meaning
A research method that checks whether people can find items in a site's structure by showing them only a text menu tree, without any visual design.
Where you use it
Often described as card sorting in reverse: card sorting helps build a structure, tree testing checks it. Participants get tasks like 'Where would you find the return policy?', and results show success rates and wrong paths.
Examples
Tree testing showed that 70% of people looked for invoices under 'Account', not 'Billing'.
Let's do tree testing on the new menu before we design any screens.
Common phrases
run tree testing · tree test task · tree testing results · reverse card sort
More about this term →Fitts's law
noun
Meaning
A principle stating that the time needed to reach a target depends on its distance and size: bigger and closer targets are faster and easier to hit.
Where you use it
Described by psychologist Paul Fitts in 1954. Designers use it to argue for larger buttons, especially on mobile, and for placing frequent actions close to where the user's finger or pointer already is.
Examples
By Fitts's law, that tiny close icon in the corner is slow and hard to tap.
Following Fitts's law, we made the 'Next' button full-width on mobile.
Usage note
The researcher's surname is Fitts, so write Fitts's law (or Fitts' law), not Fitt's law.
Common phrases
apply Fitts's law · according to Fitts's law · target size · touch target
More about this term →Machine Learning and Data Processing
artificial intelligence
noun
Meaning
The field of computer science that builds systems able to do tasks that normally need human intelligence, such as understanding language, recognizing images or making decisions.
Where you use it
Used for the whole field and, informally, for a single system ('the AI said…'). In daily work people almost always say AI; the full form appears in job titles, strategy documents, laws and formal writing.
Examples
The company is hiring three engineers for its new artificial intelligence team.
The report looks at how hospitals use artificial intelligence to plan staff shifts.
Usage note
AI is pronounced as two letters, /ˌeɪ ˈaɪ/. Because it starts with a vowel sound, say 'an AI model', not 'a AI model'.
Common phrases
AI model · AI system · AI research · AI-powered · generative AI
More about this term →machine learning
noun
Meaning
A branch of artificial intelligence in which computers learn patterns from data and use them to make predictions or decisions, instead of following rules written by hand.
Where you use it
Used as a noun and, very often, before another noun: a machine learning model, engineer or pipeline. In speech and writing it is shortened to ML, and 'AI/ML' is common in job ads.
Examples
She moved from backend development to machine learning two years ago.
We use machine learning to flag suspicious transactions before they are approved.
Usage note
'Learning' here is uncountable: say 'machine learning is', never 'a machine learning' or 'machine learnings'.
Common phrases
machine learning model · machine learning engineer · machine learning pipeline · apply machine learning
More about this term →model
noun
Meaning
The result of training: a program with learned parameters that takes an input, such as an image or a sentence, and produces an output, such as a label, a number or a text.
Where you use it
The central word of the field. You train, evaluate, deploy, serve and retrain a model; models can be small or large, open or closed, and they have versions. Informally people also say 'the model thinks…' or 'the model got it wrong'.
Examples
The new model is twice as fast but slightly less accurate on long documents.
Which model version is running in production right now?
Usage note
The US spelling is 'modeling' and the UK spelling is 'modelling'; ML libraries and papers mostly use the US form.
Common phrases
train a model · deploy a model · model version · model performance · pretrained model
More about this term →dataset
noun
Meaning
A collection of examples, such as images, texts or table rows, gathered and organized for training, testing or analyzing a model.
Where you use it
Countable: one dataset, two datasets. Engineers collect, clean, split, label and version datasets, and well-known public ones are referred to by name. In ML it is written as one word; 'data set' also appears in statistics.
Examples
We split the dataset into training, validation and test sets.
Is this dataset licensed for commercial use?
Usage note
'Data' can be pronounced /ˈdeɪtə/ or /ˈdætə/; both are correct. In tech, 'data' is usually uncountable ('the data is noisy'), so say 'a dataset', not 'a data'.
Common phrases
public dataset · split a dataset · clean a dataset · labeled dataset · dataset version
More about this term →training data
noun
Meaning
The examples a model learns from during training. Their quality, size and variety largely decide how well the model works on new data.
Where you use it
Uncountable, like 'data': 'more training data', not 'more training datas'. It is often discussed in terms of quality, coverage, licensing and privacy: what is in the training data and what is missing from it.
Examples
The model struggles with night photos because there were very few in the training data.
We removed all personal information from the training data before the first run.
Common phrases
collect training data · high-quality training data · in the training data · training data pipeline
More about this term →label
noun
Meaning
The correct answer attached to a training example, such as 'spam' for an email or 'cat' for a photo, which a supervised model learns to predict.
Where you use it
Also a verb: to label data means adding these answers, often done by paid annotators or domain experts. People talk about noisy labels, missing labels and label quality. In statistics the same idea is called the target or dependent variable.
Examples
About five percent of the labels were wrong, so we sent them back for review.
Two radiologists labeled each scan independently.
Usage note
US spelling keeps one l ('labeled', 'labeling'); UK spelling doubles it ('labelled', 'labelling').
Common phrases
noisy labels · label data · class label · label quality
More about this term →train
verb
Meaning
To teach a model by showing it many examples and repeatedly adjusting its parameters so that its outputs get closer to the correct answers.
Where you use it
You train a model on data ('train it on medical images'), for a number of epochs or steps, either from scratch or from a checkpoint. The noun is training: a training run, training job, training loop or training cost.
Examples
We trained the model on two million product reviews.
Training from scratch would take weeks, so we started from a public checkpoint.
Usage note
Turkish 'eğitmek' covers both 'educate' and 'train', but for models English uses only 'train'. 'Educate the model' is wrong.
Common phrases
train a model · train on data · train from scratch · training run · training loop
More about this term →supervised learning
noun
Meaning
A type of machine learning in which the model learns from examples that come with correct answers (labels), so that it can predict the answer for new examples.
Where you use it
Used to describe both a type of problem and a method. Classification and regression are the two main supervised tasks. In practice the biggest cost is getting enough labeled data.
Examples
Fraud detection is usually framed as a supervised learning problem, with past fraud cases as labels.
Without labeled data, we can't use supervised learning here.
Common phrases
supervised learning task · supervised learning problem · supervised model · self-supervised learning
More about this term →neural network
noun
Meaning
A model made of many connected units arranged in layers, loosely inspired by the brain. Each connection has a weight that is adjusted during training.
Where you use it
The basic building block of deep learning. People describe a network by its architecture, its number of layers and its number of parameters. In conversation it is often just 'the network' or 'the net'.
Examples
We replaced the decision tree with a small neural network and gained four points of accuracy.
The neural network has twelve layers and about 80 million parameters.
Common phrases
train a neural network · deep neural network · neural network architecture · feedforward neural network
More about this term →deep learning
noun
Meaning
The part of machine learning that uses neural networks with many layers. It powers modern speech recognition, image recognition and large language models.
Where you use it
Used for the field, its methods and its tools: deep learning frameworks, deep learning models, a deep learning course. It is usually contrasted with 'classical' or 'traditional' machine learning, which uses simpler models and hand-made features.
Examples
For tabular data, gradient-boosted trees often still beat deep learning.
Deep learning needs far more data and compute than our old pipeline did.
Common phrases
deep learning model · deep learning framework · deep learning research · apply deep learning
More about this term →loss function
noun
Meaning
A formula that measures how far a model's predictions are from the correct answers. Training tries to make this value, called the loss, as small as possible.
Where you use it
People choose, define or switch a loss function, then watch 'the loss' go down on training curves. Common examples are cross-entropy for classification and mean squared error for regression.
Examples
We switched to a weighted loss function because the model was ignoring the rare class.
Which loss function did you use for the ranking model?
Usage note
'Loss' rhymes with 'boss' (/lɔːs/); it is not pronounced like 'lose' (/luːz/).
Common phrases
choose a loss function · minimize the loss · training loss · cross-entropy loss
More about this term →overfitting
noun
Meaning
A problem in which a model learns its training data too closely, including noise and accidents, so it scores well on that data but poorly on new data.
Where you use it
A classic warning sign is training loss going down while validation loss goes up. The verb is overfit: a model overfits, or overfits to the training set. Engineers fight it with more data, regularization, dropout and early stopping.
Examples
Validation loss started rising after epoch 8, a clear sign of overfitting.
With only 300 examples, overfitting is almost guaranteed.
Common phrases
reduce overfitting · prevent overfitting · signs of overfitting · overfit to the training data
More about this term →test set
noun
Meaning
A part of the data kept aside and used only at the end, to measure how well a finished model works on examples it has never seen.
Where you use it
Kept separate from the training and validation sets. The golden rule is never to tune anything on the test set; otherwise the final score looks better than the real performance. Also called the held-out set or test split.
Examples
We ran the model on the test set only once, after all tuning was finished.
Some test set examples were also in the training data, so the score isn't reliable.
Common phrases
held-out test set · evaluate on the test set · test set accuracy · test split
More about this term →accuracy
noun
Meaning
The share of a model's predictions that are correct, out of all its predictions. A model with 90% accuracy is right on nine of every ten examples.
Where you use it
The simplest classification metric and often the first one reported. It can mislead when classes are imbalanced: if 99% of emails are not spam, a model that never says 'spam' still reaches 99% accuracy.
Examples
Accuracy went up from 87 to 91 percent after we cleaned the labels.
Don't report only accuracy here; the classes are very imbalanced.
Usage note
In everyday English 'accuracy' and 'precision' are near synonyms, but in ML they are two different metrics.
Common phrases
improve accuracy · test accuracy · high accuracy · top-1 accuracy
More about this term →inference
noun
Meaning
The stage in which a trained model is used to produce outputs for new inputs, for example answering a user's question or classifying a new photo.
Where you use it
Contrasted with training. Engineers run inference, speed it up and talk about inference cost, inference servers and batch versus real-time inference. It comes up constantly in infrastructure and budget discussions.
Examples
Now that the app has a million users, inference costs us more than training.
We run inference on CPUs at night and on GPUs during peak hours.
Usage note
In statistics and logic, inference means drawing conclusions from evidence. In ML engineering it usually just means running the model.
Common phrases
run inference · inference cost · inference speed · real-time inference · batch inference
More about this term →LLM
abbreviation
Meaning
Short for large language model: a very large neural network trained on huge amounts of text to predict the next token, which lets it answer questions, write, summarize and code.
Where you use it
Much more common than the full form in everyday talk. Engineers call an LLM through an API, host an open model themselves or build 'LLM apps'. The plural is LLMs, and you say 'an LLM' because the letter L starts with a vowel sound.
Examples
We use an LLM to draft replies, and a team member approves each one before it goes out.
Open-weight LLMs have become good enough for most of our internal tasks.
Common phrases
call an LLM · open-weight LLM · LLM application · LLM provider · fine-tune an LLM
More about this term →prompt
noun
Meaning
The text or other input given to a generative model to tell it what to do, such as a question, an instruction or a few examples followed by a task.
Where you use it
Engineers write, test, tweak and version prompts, often as templates with variables. It is also a verb: to prompt a model. In Turkish teams 'prompt' is the everyday word; 'istem' appears mostly in formal texts.
Examples
Adding two examples to the prompt fixed most of the formatting errors.
We keep every prompt in version control, just like code.
Usage note
In everyday English 'prompt' can also be an adjective meaning quick ('a prompt reply'), or a message asking the user to do something ('a login prompt').
Common phrases
write a prompt · prompt template · user prompt · prompt the model
More about this term →fine-tuning
noun
Meaning
Extra training of an already pretrained model on a smaller, task-specific dataset, so that it does better on that task or follows a particular style.
Where you use it
The noun and adjective take a hyphen (fine-tuning, a fine-tuned model); the verb is fine-tune. Teams compare fine-tuning with prompting and RAG, because it costs more but can give more consistent results.
Examples
Fine-tuning on 5,000 support tickets taught the model our tone of voice.
Before we try fine-tuning, let's see how far we get with better prompts.
Common phrases
fine-tuning data · full fine-tuning · fine-tuning job · fine-tune a model
More about this term →bias
noun
Meaning
Systematic unfairness in a model's outputs, for example working worse for some groups of people, usually caused by skewed data or design choices.
Where you use it
Teams detect, measure, audit and reduce bias, and describe it in model cards. Common phrases are 'gender bias', 'data bias' and 'bias in hiring models'. In technical papers the word can also have a purely mathematical meaning, so check the context.
Examples
The audit found a clear bias: the model rejected older applicants more often.
We added more speakers with regional accents to reduce bias in speech recognition.
Usage note
'Bias' also has neutral technical meanings in ML: the extra learned number in each neuron (the bias term) and one kind of error in the bias-variance trade-off. Neither is about fairness.
Common phrases
reduce bias · measure bias · gender bias · data bias · bias audit
More about this term →classification
noun
Meaning
A task in which a model puts each input into one of a fixed set of categories, for example deciding whether an email is spam or which product category a photo shows.
Where you use it
Classification can be binary (two classes), multi-class (one of many) or multi-label (several at once). The model is called a classifier, and its quality is measured with accuracy, precision, recall and F1 score.
Examples
This is a multi-label classification problem: one ticket can be about billing and delivery at the same time.
For simple classification tasks, a small fine-tuned model is cheaper than an LLM.
Common phrases
binary classification · multi-class classification · image classification · text classification · classification model
More about this term →natural language processing
noun
Meaning
The area of AI that deals with human language, building systems that read, understand, translate, summarize or generate text and speech.
Where you use it
Almost always shortened to NLP in speech and job titles (NLP engineer, NLP team). Classic NLP tasks include sentiment analysis, named entity recognition and machine translation; since LLMs, many of them are handled by one general model.
Examples
Turkish is harder for natural language processing because one word can carry many suffixes.
She wrote her master's thesis on natural language processing for legal documents.
Usage note
Don't confuse it with neuro-linguistic programming, a self-help method that is also shortened to NLP.
Common phrases
NLP engineer · NLP task · NLP pipeline · classic NLP
More about this term →computer vision
noun
Meaning
The area of AI that teaches computers to understand images and video, for example recognizing objects, reading text in photos or tracking movement.
Where you use it
Used for the field and in job titles (computer vision engineer). Typical tasks are image classification, object detection, segmentation and tracking, in areas such as manufacturing, medical imaging, retail and self-driving cars. The short form CV can be confused with CV meaning a résumé.
Examples
Our computer vision system checks every bottle on the line for cracks.
Most of our computer vision models run on the cameras themselves, not in the cloud.
Common phrases
computer vision model · computer vision engineer · computer vision task · computer vision pipeline
More about this term →unsupervised learning
noun
Meaning
A type of machine learning that finds structure in data without correct answers (labels), for example grouping similar customers or spotting unusual events.
Where you use it
Typical methods are clustering, dimensionality reduction and anomaly detection. It helps when labels are expensive or impossible to get. A related idea, self-supervised learning, creates its own labels from the data and is how LLMs are pretrained.
Examples
We used unsupervised learning to group support tickets before anyone had labeled them.
Unsupervised learning flagged a group of payments that later turned out to be fraud.
Common phrases
unsupervised learning method · unsupervised model · unsupervised anomaly detection · self-supervised learning
More about this term →validation set
noun
Meaning
A part of the data kept out of training and used during development to compare models, tune settings and decide when to stop training.
Where you use it
It sits between the training set and the test set. Engineers check validation loss after each epoch and choose hyperparameters by validation score. Also called the dev set, especially in NLP.
Examples
We picked the learning rate that gave the best score on the validation set.
The validation set is too small: one wrong prediction changes accuracy by two points.
Common phrases
validation loss · validation accuracy · hold out a validation set · validation split
More about this term →epoch
noun
Meaning
One complete pass of the training process through the whole training dataset. Models are usually trained for several epochs.
Where you use it
Used with numbers: train for 10 epochs, after epoch 3, loss per epoch. Very large models may see their data only once, so people count steps or tokens instead. In Turkish teams the English word is used almost always.
Examples
Validation accuracy stopped improving after the fifth epoch.
Each epoch takes about forty minutes on four GPUs.
Usage note
American English says /ˈepək/ (EP-uk); British English usually says /ˈiːpɒk/ (EE-pok). Both are understood.
Common phrases
train for ten epochs · per epoch · number of epochs · after each epoch
More about this term →learning rate
noun
Meaning
A hyperparameter that controls how big a step the model takes each time it updates its parameters. Too high makes training unstable; too low makes it very slow.
Where you use it
Usually the first hyperparameter people tune. It is often changed during training by a learning rate schedule, with a short warm-up at the start and a gradual decay later. Written as LR in notes and code, for example 'LR of 3e-4'.
Examples
The loss exploded after a few hundred steps, so we lowered the learning rate.
We increase the learning rate slowly during the first 1,000 steps as a warm-up.
Common phrases
lower the learning rate · learning rate schedule · learning rate warm-up · initial learning rate
More about this term →hyperparameter
noun
Meaning
A setting chosen by engineers before training rather than learned from data, such as the learning rate, the batch size or the number of layers.
Where you use it
You tune, search and log hyperparameters. Hyperparameter tuning or hyperparameter search means trying many combinations and keeping the best one by validation score. In code and chat people often write 'hparams'.
Examples
We ran a hyperparameter search over 40 configurations overnight.
Please log all hyperparameters with each run so we can reproduce the results.
Common phrases
hyperparameter tuning · hyperparameter search · tune hyperparameters · default hyperparameters
More about this term →batch size
noun
Meaning
The number of training examples a model processes together before it updates its parameters once.
Where you use it
Limited mainly by GPU memory. Bigger batches use the hardware more efficiently but may need a different learning rate. When memory runs out, engineers reduce the batch size or use gradient accumulation.
Examples
We got an out-of-memory error, so I cut the batch size from 64 to 32.
With a larger batch size, each epoch finishes faster.
Common phrases
increase the batch size · effective batch size · mini-batch · batch size of 32
More about this term →gradient descent
noun
Meaning
The basic optimization method behind training: it repeatedly changes each parameter a little in the direction that reduces the loss the most.
Where you use it
In practice people use variants: stochastic gradient descent (SGD), which works on small random batches, and optimizers such as Adam that build on it. The gradients themselves are computed by backpropagation.
Examples
Gradient descent was barely moving because the learning rate was far too small.
In the interview they asked me to explain gradient descent on a whiteboard.
Common phrases
stochastic gradient descent · run gradient descent · gradient descent step · mini-batch gradient descent
More about this term →underfitting
noun
Meaning
A problem in which a model is too simple or trained too little to capture the patterns in the data, so it performs poorly even on the training data.
Where you use it
The opposite of overfitting. The usual sign is high training loss and high validation loss together. Typical fixes are a bigger model, better features, longer training or less regularization. The verb is underfit.
Examples
Both losses are still high, so this looks like underfitting, not overfitting.
A linear model leads to underfitting on data this complex.
Common phrases
signs of underfitting · fix underfitting · underfit the data
More about this term →precision
noun
Meaning
Of all the cases a model marked as positive, the share that really were positive. High precision means few false alarms.
Where you use it
Usually discussed together with recall, because improving one often lowers the other. Precision matters most when a false positive is costly, for example blocking a real customer's payment as fraud.
Examples
The spam filter has 98% precision, so almost nothing it blocks is real mail.
We raised the threshold to improve precision, but recall dropped.
Usage note
Turkish sources are not consistent: precision is usually 'kesinlik', but some texts call it 'hassasiyet', a word others use for recall. Many teams simply use the English terms to avoid confusion.
Common phrases
high precision · precision and recall · improve precision · precision at k
More about this term →recall
noun
Meaning
Of all the cases that really were positive, the share that the model managed to find. High recall means few missed cases.
Where you use it
Recall matters most when missing a case is costly, for example failing to flag a tumor on a scan. In medicine the same measure is called sensitivity. It is reported together with precision or combined with it in the F1 score.
Examples
For the screening model, we care more about recall than precision.
Recall on the rare class is only 40%, so most of those cases are missed.
Usage note
In everyday English, recall means remembering something, or a company calling back faulty products. As a metric, the stress can fall on either syllable: /ˈriːkɔːl/ or /rɪˈkɔːl/.
Common phrases
high recall · recall rate · precision and recall · recall at k
More about this term →F1 score
noun
Meaning
A single number that combines precision and recall by taking their harmonic mean. It is high only when both precision and recall are high.
Where you use it
A standard classification metric, especially with imbalanced classes where accuracy misleads. With many classes it is averaged in different ways: macro F1 treats every class equally, while micro F1 treats every example equally.
Examples
The new model raised the F1 score from 0.71 to 0.78.
Please also report the macro-averaged F1 score; the small classes matter to the client.
Common phrases
report the F1 score · macro F1 · micro F1 · F1 on the test set
More about this term →benchmark
noun
Meaning
A standard set of tasks and data used to measure models in the same way, so that results from different teams can be compared fairly.
Where you use it
Models are evaluated on, scored on and compared across benchmarks, and a new best result tops a benchmark or a leaderboard. It is also a verb: to benchmark a model. Engineers often warn that high benchmark scores don't always mean good real-world performance.
Examples
It scores well on public benchmarks but struggles with our customers' questions.
We built an internal benchmark from 500 real support conversations.
Common phrases
public benchmark · internal benchmark · benchmark score · top a benchmark · benchmark results
More about this term →transformer
noun
Meaning
A neural network architecture, introduced in 2017, that uses attention to process all parts of a sequence in parallel. It is the basis of today's LLMs.
Where you use it
Used as a noun and before other nouns: a transformer model, transformer layers, a transformer-based system. Engineers talk about encoder-only, decoder-only and encoder-decoder transformers. They are used for images and audio too, not only text.
Examples
We replaced the old recurrent model with a small transformer and cut errors by a third.
Most current LLMs are decoder-only transformers.
Usage note
In electrical engineering a transformer is a device that changes voltage; in ML the word always means this architecture.
Common phrases
transformer model · transformer architecture · decoder-only transformer · transformer-based
More about this term →layer
noun
Meaning
One level of a neural network: a group of units that takes the output of the previous level, transforms it and passes the result on to the next one.
Where you use it
Networks have an input layer, hidden layers and an output layer. Engineers add, remove, freeze or unfreeze layers, and the 'deep' in deep learning refers to having many of them. Common types include dense, convolutional and attention layers.
Examples
We froze the first ten layers and trained only the top two.
Adding another hidden layer didn't help; the model just overfit faster.
Common phrases
hidden layer · output layer · freeze layers · add a layer · dense layer
More about this term →hallucination
noun
Meaning
A confident but false or made-up output from a generative model, such as an invented fact, quote, source or piece of code.
Where you use it
Teams measure, detect and reduce hallucinations, often with retrieval (RAG), citations and automatic checks. The verb is hallucinate: 'the model hallucinated a paper that doesn't exist'. Some researchers prefer the word 'confabulation'.
Examples
This answer is a hallucination: the two court cases it cites never existed.
Grounding answers in our own documents cut hallucinations by more than half.
Common phrases
reduce hallucinations · hallucination rate · detect hallucinations · the model hallucinated
More about this term →context window
noun
Meaning
The maximum amount of text, measured in tokens, that a language model can take into account at once, including the prompt, any documents and its own answer.
Where you use it
Described by its size: a 128k context window, a long context window. When a conversation grows longer than the window, the oldest parts must be cut or summarized. Also called context length.
Examples
The whole contract fits in the context window, so we don't need to split it.
Answer quality drops when we fill the last part of the context window.
Common phrases
context window size · long context window · fit in the context window · 128k context window
More about this term →generative AI
noun
Meaning
AI systems that create new content, such as text, images, audio, video or code, instead of only sorting inputs into classes or predicting values.
Where you use it
Used for both the technology and the industry around it, and shortened to GenAI in business talk. Companies discuss generative AI features, use cases and policies, while engineers usually prefer more precise words such as LLM or diffusion model.
Examples
Legal asked us to write a generative AI policy before the tool goes live.
Most of our generative AI features are built on two external LLM providers.
Common phrases
generative AI tools · generative AI policy · generative AI use case · GenAI
More about this term →embedding
noun
Meaning
A list of numbers (a vector) that represents a word, sentence, image or other item in such a way that items with similar meanings end up close to each other.
Where you use it
Usually plural. You compute, store and compare embeddings, typically with cosine similarity. They power semantic search, recommendations, clustering and RAG systems, and are kept in vector databases.
Examples
We generate an embedding for every help article and store it in the vector database.
Search results improved after we switched to a multilingual embedding model.
Common phrases
generate embeddings · embedding model · text embeddings · embedding vector · embedding space
More about this term →system prompt
noun
Meaning
Instructions given to a chat model before the conversation starts, which set its role, rules, tone and limits for all the messages that follow.
Where you use it
Written by developers, not end users, and usually hidden from them. Teams keep it in version control and test changes carefully, because a small edit can change behavior everywhere. It is contrasted with the user prompt or user message.
Examples
The system prompt tells the assistant to answer only questions about our products.
After we shortened the system prompt, responses became faster and cheaper.
Common phrases
write a system prompt · update the system prompt · system prompt instructions · leaked system prompt
More about this term →RAG
abbreviation
Meaning
Short for retrieval-augmented generation: a method in which a system first searches a document collection for relevant passages and then gives them to an LLM to write the answer.
Where you use it
Pronounced like the word 'rag'. Teams build RAG pipelines over company documents so that answers stay up to date and can cite sources without retraining the model. Quality depends heavily on retrieval: chunking, embeddings and ranking.
Examples
With RAG, the chatbot can answer questions about policies we changed last week.
Most wrong answers came from the retrieval step, not the LLM, so we fixed that part of the RAG pipeline first.
Common phrases
RAG pipeline · RAG system · build a RAG app · RAG evaluation
More about this term →annotation
noun
Meaning
The work of adding labels or notes to raw data, such as drawing boxes around objects in images or marking the topic of each sentence, so that it can be used for training or evaluation.
Where you use it
Also used for the labels themselves ('the annotations are in a JSON file'). The people who do it are annotators; teams write annotation guidelines and measure how often annotators agree. Often used as a near synonym of labeling.
Examples
Annotation of 10,000 X-rays took three doctors almost two months.
We rewrote the annotation guidelines because annotators disagreed on a third of the examples.
Common phrases
annotation guidelines · annotation tool · data annotation · inter-annotator agreement
More about this term →preprocessing
noun
Meaning
The steps that clean and prepare raw data before a model uses it, such as removing duplicates, filling in missing values, resizing images or normalizing text.
Where you use it
Engineers write preprocessing steps, scripts and pipelines. The same preprocessing must be applied during training and in production; differences between the two are a common source of bugs. The verb is preprocess.
Examples
The bug was in preprocessing: production images weren't resized the same way.
On this dataset, preprocessing takes longer than training.
Common phrases
data preprocessing · preprocessing pipeline · preprocessing step · text preprocessing
More about this term →reinforcement learning
noun
Meaning
A type of machine learning in which an agent learns by trial and error, taking actions in an environment and receiving rewards or penalties for the results.
Where you use it
Shortened to RL. Its key words are agent, environment, action, reward and policy. It is used in game-playing systems, robotics and recommendations, and in the RLHF stage that makes LLMs more helpful.
Examples
The warehouse robots were first trained with reinforcement learning in a simulator.
Designing the reward is the hardest part of any reinforcement learning project.
Common phrases
reinforcement learning agent · deep reinforcement learning · reward function · RL environment
More about this term →clustering
noun
Meaning
An unsupervised task that groups similar items together without predefined labels, for example sorting customers into segments by how they shop.
Where you use it
Common algorithms are k-means, DBSCAN and hierarchical clustering. Each group is a cluster, and items are said to 'cluster together'. It is often run on embeddings to find themes in large collections of texts or images.
Examples
Clustering the complaints revealed three main themes: delivery, refunds and app crashes.
We ran clustering on the embeddings and then named each cluster by hand.
Common phrases
k-means clustering · run clustering · cluster analysis · customer clustering
More about this term →weights
noun
Meaning
The learned numbers on the connections of a neural network, which decide how strongly each input affects the next layer. Together they hold what the model has learned.
Where you use it
Usually plural. Engineers load, save, freeze, share and download weights, and 'the weights' often means the trained model file itself. An open-weight model is one whose weights are published, even if its training data and code are not.
Examples
Download the weights from the shared drive and load them into the evaluation script.
They published the weights but not the training data.
Usage note
'Open-weight' and 'open-source' are not the same: open source normally also means that the code is shared and the license allows free use and changes.
Common phrases
model weights · load the weights · open weights · freeze the weights · pretrained weights
More about this term →activation function
noun
Meaning
A simple mathematical function applied to the output of each unit in a neural network. It adds non-linearity, which lets the network learn complex patterns.
Where you use it
Common examples are ReLU, sigmoid, tanh and GELU. Without activation functions, even a deep network would behave like a single linear layer. Engineers choose or swap the activation function when they design an architecture.
Examples
Try ReLU as the activation function in the hidden layers and sigmoid at the output.
Which activation function does this transformer use?
Common phrases
ReLU activation function · choose an activation function · non-linear activation function · output activation
More about this term →backpropagation
noun
Meaning
The method that calculates how much each weight in a neural network contributed to the error, by passing the error backward from the output layer toward the input.
Where you use it
Often shortened to backprop. It works together with gradient descent: backpropagation computes the gradients, then the optimizer updates the weights. Frameworks do it automatically, so engineers rarely write it by hand outside courses and interviews.
Examples
During backpropagation, the gradients in the early layers became almost zero.
Our deep learning course spends a whole week on backpropagation.
Common phrases
backpropagation algorithm · run backpropagation · backpropagation through time · compute gradients
More about this term →attention
noun
Meaning
A mechanism that lets a model decide which other parts of its input are most relevant to each part, and give those parts more weight when it builds its output.
Where you use it
The core idea of the transformer. Engineers talk about self-attention, attention heads, attention layers and attention maps. Because its cost grows quickly with input length, a lot of research aims to make attention faster.
Examples
The attention maps suggest the model focuses on the drug name when it answers.
Each layer has 16 attention heads.
Usage note
In general English, 'pay attention' means to focus on something. In ML, attention is a precise technical mechanism, not a human-like ability.
Common phrases
self-attention · attention head · attention layer · attention mechanism · attention map
More about this term →ground truth
noun
Meaning
The correct, verified answers that a model's outputs are compared with, usually created by careful human labeling or by real measurements.
Where you use it
Used as a noun and before other nouns: ground truth labels, ground truth data. People also question whether the ground truth itself is reliable, since human labels can be wrong or disagree. Common in evaluation, computer vision and remote sensing.
Examples
For this test we used biopsy results as the ground truth.
The model disagrees with the ground truth on 6% of the images, and some of those labels are wrong.
Common phrases
ground truth labels · ground truth data · compare against the ground truth · establish the ground truth
More about this term →data augmentation
noun
Meaning
Creating extra training examples by making small changes to existing ones, such as flipping or cropping images or rephrasing sentences, so that a model sees more variety.
Where you use it
Very common in computer vision and speech, and increasingly used in NLP. It helps reduce overfitting when data is limited. Engineers must check that a change keeps the label correct: a rotated '6' may turn into a '9'.
Examples
Data augmentation with random crops and color changes added three points of accuracy.
We turned off data augmentation for the validation set.
Common phrases
data augmentation techniques · apply data augmentation · image augmentation · augmentation pipeline
More about this term →confusion matrix
noun
Meaning
A table that shows, for each true class, how many examples the model assigned to each class, so you can see at a glance which classes it mixes up.
Where you use it
For two classes it has four cells: true positives, false positives, true negatives and false negatives. Precision and recall are calculated from these counts. Teams plot it as a heatmap and look at it in almost every model review.
Examples
The confusion matrix shows that the model often mixes up wolves and huskies.
Can you add a confusion matrix for each language to the report?
Common phrases
plot a confusion matrix · normalized confusion matrix · false positives · false negatives
More about this term →cross-validation
noun
Meaning
An evaluation method that splits the data into several parts, trains on all but one and tests on the remaining part, repeating until every part has been used for testing.
Where you use it
The most common form is k-fold cross-validation, for example 5-fold. It gives a more reliable estimate than a single split, especially on small datasets, but needs k times as much training. It is less common with very large models.
Examples
With only 800 patients, we used 10-fold cross-validation instead of a fixed test set.
The scores vary a lot across the cross-validation folds.
Common phrases
k-fold cross-validation · 5-fold cross-validation · cross-validation score · cross-validation folds
More about this term →