Learn AI /Data and Infrastructure /Metadata
Metadata
- Starter
- Data and Infrastructure
- Programme & data teams
Back to the milestone you came from
In five seconds
What it is
Information that describes other data, such as who created it, when, where, in what format, and under what rules it can be used.
In plain words
Metadata is the information attached to a file or dataset that tells you what it is, where it came from, and how it should be handled.
An everyday example
A city office stores consultation papers in archive boxes. The papers inside are the main content, but staff first read the labels on the outside: project name, year, language, department, and whether access is restricted. Those labels help them find the right box quickly and stop them opening boxes they are not allowed to use. If a label is wrong or missing, staff may waste time, use old material, or handle restricted papers badly.
A box label is separate from the papers inside, but real metadata is not always outside or obvious. It may be stored inside the file, in a database, or added by another system. Also, metadata can be generated automatically and can change over time, while a paper label is usually fixed until a person rewrites it.
Remember thisMetadata describes a data resource so people and systems can find it, interpret it, trace its origin and apply access, version and retention rules.
Understand it
Understand it
Why it matters
Metadata affects everyday decisions. It helps people find the right file, judge whether it is current, filter large collections, and apply rules about access, retention, and sharing. If metadata is missing or wrong, teams may use outdated material, misclassify documents, or expose sensitive information. For AI work, metadata also shapes quality checks, sampling, evaluation, and governance because it tells you where data came from, what it contains, and what limits apply. A non-specialist does not need to design metadata systems, but does need to recognise that good metadata makes data usable and safer, while bad metadata creates delay, confusion, and risk.
How metadata helps people handle data
Show how descriptive details attached to data make searching, organising, and controlling access possible.
01
1. Data item
Start with a piece of content such as a report, photo, spreadsheet, or audio file.
Draw one central file card labelled 'Report' with a small preview block but no detailed text.
02
2. Descriptive details added
Attach fields like title, date, language, author, format, topic, and access level.
Place small tagged chips around the file card: 'date: 2026-03-12', 'language: Arabic', 'topic: water', 'access: internal'.
03
3. Metadata is read
A person or system reads those fields instead of scanning every line of the content first.
Show arrows from the tagged chips to a search icon, a folder icon, and a shield icon.
04
4. Practical outcome
The data can be found, sorted, filtered, and protected more effectively.
Show three output panels labelled 'find quickly', 'organise correctly', and 'share safely'.
A left-to-right flow with four stages. Stage 1 shows a data item. Stage 2 adds descriptive fields around it. Stage 3 shows systems and people reading those fields. Stage 4 shows the results: faster search, better organisation, and safer handling. Read from left to right.
Hovering or tapping each metadata chip highlights one output it supports, such as date helping sorting or access level helping protection.
Describe this without the diagram
The diagram shows a simple flow from a data item to useful actions. First there is one content item, shown as a report. Next, several metadata fields are attached to it: date, language, topic, and access level. Then people and software read those fields. Because of that, three outcomes become possible: finding the item quickly, organising it correctly, and sharing it more safely. The main point is that metadata does not replace the content. It describes the content so it can be managed better.
With reduced motion: A static four-panel layout shows the same sequence without animation: data item, metadata fields, reading by people and systems, then outcomes of search, organisation, and safer handling.
Metadata is data about data. It does not usually contain the main content itself. Instead, it describes the content so people and systems can find it, sort it, use it correctly, and manage it safely.
What goes in can be many kinds of descriptive details: a file name, date created, language, location, author, subject tag, document type, data format, or access level. For a photo, metadata might include the time it was taken and the device used. For a spreadsheet, it might include column names, units, and who last updated it.
What happens next is organisation and control. Software can read metadata to group similar items, search faster, apply rules, track versions, or warn that a dataset is restricted. People also use it to judge whether data is relevant, recent, complete, or suitable for a task.
What comes out is better handling of the underlying data: easier discovery, clearer context, and safer sharing. But metadata can also create risk. Even if the main content is hidden, metadata may still reveal sensitive details such as names, locations, dates, or relationships, so it needs governance and care.
How “Labels on archive boxes” maps onto the real thing
| In the story | In the AI system |
|---|---|
| Labels on the outside of each box | Metadata that describes a file or dataset |
| Project name, year, language, department | Fields such as topic, date, language, source, and owner |
| Restricted access mark | Permission or sensitivity metadata |
| Choosing the right box without reading every page | Searching, filtering, and managing data using metadata |
A worked example
A reporting officer in a regional bureau needs the latest cleared monitoring report to prepare talking points before close of business.
Starting from: A folder contains three similarly named files: an editable draft, a revised draft and a PDF. Their metadata records creator, creation time, reporting period, version, format and clearance status.
- The officer filters the folder by the required reporting period and document type.
- The metadata shows that the editable files are drafts, while the PDF is the later version cleared by the chief of section.
- The officer checks the provenance and version history to confirm that the PDF derives from the revised draft.
- The cleared PDF is used for the talking points, while the drafts remain on the file under their applicable retention rules.
Result: The officer selects the authoritative version without opening and comparing every file. The decision rests on recorded provenance, version and clearance metadata, not on an optimistic filename.
In a United Nations workplace
A records officer in a liaison office prepares a handover note for an incoming colleague. For each file cited, the officer checks metadata including the creator, document type, language, version, clearance status, access level and retention action, rather than relying on filenames such as “final” and “final revised”.
The metadata helps the incoming colleague find the current aide-memoire, distinguish cleared material from drafts, trace where a document came from and handle it under the correct access and retention rules.
Before you try thisMetadata can itself disclose identities, locations, working relationships or the existence of restricted material. Personal, security-sensitive or otherwise confidential metadata must not be copied into an external tool unless that tool is approved for the material. An access label is also not a substitute for properly configured permissions.
Where you may hear it
- A records handover in a liaison office
- A vendor demo for a document management system
- An information management meeting on a data catalogue
- A clearance discussion about which file is authoritative
- An internal audit interview on access and retention controls
Avoid confusion
Avoid confusion
Often confused with
Data
Recorded observations, measurements or examples that people or systems can store, examine and use for a purpose.
Structured data
Information organised into predefined fields, often in rows and columns, so software can process it predictably.
Annotation and data labelling
People or software tools add task-relevant tags, notes or boundaries to data for model training, testing, analysis or monitoring.
Data governance
Data governance is the set of rules, roles and checks that control how data is collected, stored, used, shared and corrected over time.
Common misconceptions
Common misconception: Metadata is just technical clutter that only IT staff need.
Correction: Many non-technical tasks depend on it. Dates, languages, authors, version numbers, and access labels help programme staff, researchers, librarians, and managers find the right material and handle it properly. Without metadata, even simple filing and review becomes slower and riskier.
Common misconception: If the main content is removed, metadata is harmless.
Correction: Not always. Metadata can still reveal sensitive facts, such as who created a file, when it was sent, where a photo was taken, or which people or organisations are connected to it. That is why metadata also needs protection and governance.
Common misconception: Metadata and data are two completely separate things.
Correction: They are different roles, not always different objects. Metadata describes other data, but it is still data in its own right. In one context, a date field may be metadata about a document; in another, that same date can be analysed as ordinary data.
Quick check
Which option is the best example of metadata for a report file?
Metadata describes the report rather than being the report's main content. Fields like title, date, language, and access level help people and systems find, sort, and govern the file.
Go deeper
Go deeper
The simple version
Metadata is the information attached to a file or dataset that tells you what it is, where it came from, and how it should be handled.
The technical version
Metadata is structured or semi-structured descriptive information about a data object, dataset, or information resource. Common metadata fields include provenance, meaning where it came from; schema, meaning how fields are organised; timestamps; format; ownership; version; quality indicators; and access or retention rules. Systems use these fields for indexing, search, filtering, lineage tracking, interoperability, lifecycle management, and policy enforcement. Metadata can be embedded in a file, stored in a sidecar file, or managed in a separate catalogue. It is often confused with the primary content, but the key distinction is functional: metadata describes, organises, or governs another data asset, even though it may itself later be analysed as data.
Reviewed external resources
Primary sourceAI Risk Management Framework (opens in a new tab)NIST
The NIST AI Risk Management Framework helps readers see how documentation, traceability, and governance-related metadata support safer AI practices, and it suits practitioners and managers.
Intermediatelink checked 2026-07-31
Primary sourceOECD AI Principles (opens in a new tab)OECD
The OECD AI Principles help policy and programme readers connect metadata to accountability, transparency, and responsible data handling.
Intermediatelink checked 2026-07-31
EncyclopaediaMetadata (opens in a new tab)Wikipedia
Wikipedia gives a broad introductory overview of metadata and suits readers who want a quick general starting point.
Intermediatelink checked 2026-07-31
VideoVideo explanations: Metadata (opens in a new tab)YouTube search
The YouTube search results page offers a range of video explanations of metadata for different learning styles, and it suits readers who prefer visual or spoken introductions.
search results, not one pagelink checked 2026-07-31
DiscussionCommunity discussion: Metadata (opens in a new tab)Reddit search
The Reddit search results page shows community questions and discussions about metadata, and it suits readers who want to see common confusions and practical examples.
search results, not one pagelink checked 2026-07-31
Where this came from
Drafted with a language model through an offline pipeline, checked against this site's schema, and every external link above fetched to see whether it answers. Treat it as a starting point and check anything that matters against the primary sources. How this site is built and what that means.
The one thing to remember
Metadata describes a data resource so people and systems can find it, interpret it, trace its origin and apply access, version and retention rules.