Frequently Asked Questions

Tagging.tech explores how people and machines describe, categorize, and organize digital content.

Whether you are completely new to metadata and keywording or have spent years managing digital assets, this FAQ explains the concepts, technologies, and services discussed throughout Tagging.tech.

About Tagging.tech

What is Tagging.tech?

Tagging.tech is a resource focused on the people, processes and technologies used to tag audio, images and video.

The site explores how humans and machines apply keywords, metadata, and descriptive information to digital content. Through interviews and resources, Tagging.tech examines human keywording services, image recognition, video recognition, artificial intelligence, computer vision and crowdsourcing.

Who is Tagging.tech for?

Tagging.tech is for anyone interested in making digital content easier to organize, find and use.

This includes Digital Asset Management professionals, metadata specialists, archivists, librarians, content operations teams, marketers, photographers, videographers, technology professionals and organizations managing growing collections of digital assets.

You do not need to be a metadata expert to benefit from the site.

Who created Tagging.tech?

Tagging.tech was created by Henrik de Gyor, a consultant, author, speaker, and podcaster with extensive experience in Digital Asset Management, metadata, digital content operations, and digital transformations.

The site was developed to explore how people and technology describe digital content and how tagging continues to affect the way organizations manage information.

Why does tagging matter?

Digital content has limited value when people cannot find it.

Tagging adds descriptive information to audio, video, text, graphics and photographs so content can be searched, filtered, organized, used, reused, repurposed and actionable.

Good tagging can improve findability, improve relevant results, minimize scrolling, reduce duplicate work, and help organizations get more value from their digital assets.

Tagging and Keywording

What is tagging?

Tagging is the process of adding descriptive words, phrases or other information to digital content.

For example, a photograph of a red Porsche driving on a mountain road could include tags such as Porsche, sports car, red car, mountain, road and driving, among other specific attributes, like the model, trim level, location, and who is driving.

These tags help people and systems understand what the content contains.

What is keywording?

Keywording is the process of assigning descriptive terms to content.

The terms are selected to represent the subject, concept, location, activity, person, product, or other characteristics associated with an asset.

Keywording is commonly used with photographs, video, audio, documents, and other digital assets.

Are tagging and keywording the same thing?

The terms are often used interchangeably.

Keywording usually refers specifically to descriptive words or phrases assigned to content. Tagging can have a broader meaning and may include keywords, labels, categories, or other metadata values.

The exact terminology often depends on the system, industry, and organization.

What is metadata?

Metadata is information that describes other information.

For a photograph, metadata might include the photographer’s name, copyright information, creation date, location, keywords, and a description of the image.

For a video, metadata might include the title, duration, speakers, subjects, and topics covered.

Tagging and keywording are forms of metadata.

What makes a good keyword?

A good keyword accurately describes the content and helps someone find it.

Keywords should be relevant, understandable, and applied consistently.

Adding more keywords does not automatically improve search results. Irrelevant, vague, or inconsistent keywords can make finding content more difficult.

Can you add too many keywords?

Yes.

Adding every possible word associated with an asset can create metadata noise.

The goal is not to generate the longest possible list of keywords. The goal is to use relevant terms that improve findability and accurately describe the asset.

What is a controlled vocabulary?

A controlled vocabulary is an approved list of terms used to describe and categorize information.

For example, an organization may decide to use the term “automobile” instead of allowing automobile, auto, car and motorcar to be used inconsistently.

Controlled vocabularies can improve metadata consistency and search results.

What is a taxonomy?

A taxonomy organizes terms into categories and relationships.

For example:

Vehicle

Automobile

Sports Car

Porsche

A taxonomy can help users browse information and can guide how metadata is applied.

What is the difference between a taxonomy and a controlled vocabulary?

A controlled vocabulary defines the approved terms that should be used.

A taxonomy organizes those terms into a structure.

Organizations often use both together.

Human Keywording

What is human keywording?

Human keywording involves a person reviewing an asset and assigning descriptive terms to it.

The keyworder may identify objects, people, activities, emotions, locations, concepts, and other information represented in the content.

What are the advantages of human keywording?

Humans can often recognize context, nuance and meaning that automated systems may miss.

A person may understand humor, symbolism, cultural context or the intent behind an image.

Human keyworders can also follow detailed organizational rules and controlled vocabularies.

What are the disadvantages of human keywording?

Human keywording can require considerable time and resources.

Different people may also describe the same asset differently.

Without clear metadata standards, training, and quality control, keywording can become inconsistent.

What is a professional keywording service?

A professional keywording service provides people who specialize in reviewing and describing digital content.

Organizations may use these services when they have large collections of images or other media that need metadata.

Keywording services can be particularly useful for content migrations, archives, and large metadata improvement projects.

What is crowdsourced tagging?

Crowdsourced tagging uses a group of people to describe or categorize content.

The participants may be employees, customers, volunteers, or workers using a crowdsourcing platform.

Crowdsourcing can process large amounts of content, but requires careful quality control.

Can different people tag the same image differently?

Absolutely.

One person might tag an image “car.” Another might use “Porsche 911.” Someone else might focus on the setting and use “mountain road.”

All three descriptions may be technically correct.

This is one reason metadata standards and tagging guidelines are important.

Machine Tagging and Image Recognition

What is automated tagging?

Automated tagging uses software to analyze digital content and generate descriptive information.

Depending on the technology, the system may identify objects, faces, scenes, text or other characteristics.

What is image recognition?

Image recognition uses computer technology to analyze and identify visual information within an image.

A system might recognize a dog, automobile, building, or person.

Some image recognition systems can identify specific products, logos, or landmarks.

What is computer vision?

Computer vision is a field of technology focused on helping computers analyze and interpret visual information.

Image recognition is one application of computer vision.

Computer vision technologies may be used for object detection, facial recognition, scene analysis, and other visual processing tasks.

How accurate is machine-generated tagging?

Accuracy varies considerably.

The technology, training data, image/audio quality and subject matter expertise shared can all affect results.

Automated tagging may work very well with common objects, but can struggle with specialized products, unusual situations, or concepts that require contextual understanding if it has not been provided as training, and often requires a human in the loop to point it in the right direction to produce desired results and correct as needed.

Organizations should test automated tagging technologies using their own content before assuming the results will meet their requirements.

Can machines understand the meaning of an image?

Machines can identify patterns and characteristics within images.

Understanding context and meaning can be more difficult.

For example, a system may recognize two people, a table and food. A human may immediately understand that the image shows a family celebrating a birthday.

The distinction between identifying objects and understanding the meaning of a scene is important.

What is object recognition?

Object recognition identifies objects within an image or video.

Examples might include automobile, bicycle, dog, building, or computer.

Some systems can identify more specialized objects depending on how the technology was trained.

What is facial recognition?

Facial recognition uses technology to analyze facial characteristics and potentially identify a person.

Facial recognition is different from face detection.

Face detection determines that a face is present. Facial recognition attempts to determine whose face it is.

Organizations considering facial recognition should carefully evaluate privacy, legal, and governance requirements.

Can automated tagging replace human keywording?

Sometimes automation can dramatically reduce the amount of human work required.

That does not mean human review is always unnecessary.

Many organizations benefit from combining machine-generated tags with human review. Technology can process content quickly while people validate important information and add context.

The right approach depends on the content, scale and business requirements.

What is human in the loop tagging?

Human in the loop tagging combines automation with human review.

A machine may generate suggested tags and a person confirms, removes or adds terms.

This approach can combine the speed of automated processing with human understanding and quality control.

Video and Audio Tagging

Can video be automatically tagged?

Yes.

Video recognition technologies can analyze frames, objects, scenes, faces and other visual information.

Some systems also analyze speech and text contained within a video.

Because video contains large amounts of information, automated analysis can be particularly useful when working with extensive video collections.

How is video tagging different from image tagging?

An image represents a single visual asset.

A video changes over time.

Objects, people and topics may appear at different points in a video. Effective video tagging may therefore include time based metadata that identifies when something appears.

What is time based metadata?

Time-based metadata connects descriptive information to a specific point or segment within audio or video.

For example, a one hour interview might discuss metadata at 12 minutes and image recognition at 35 minutes.

Time-based metadata can help someone navigate directly to the relevant section.

Can audio be tagged automatically?

Yes.

Technologies can analyze spoken language, music and other audio characteristics.

Speech to text systems can create transcripts that can then be searched or analyzed for topics and keywords.

Why are transcripts useful for tagging?

A transcript converts spoken words into searchable text.

Keywords, names, topics and phrases can be identified within the transcript.

For interviews, podcasts and video recordings, transcripts can significantly improve findability.

Metadata Strategy

What is a metadata strategy?

A metadata strategy defines how an organization will describe, organize and manage information.

It should address what metadata is required, who creates it, how it is governed and how it supports business needs.

A successful metadata strategy starts with understanding how people need to find and use content.

Why do organizations struggle with metadata?

Metadata problems are rarely caused by a lack of fields.

Common challenges include inconsistent terminology, unclear ownership, limited governance, poor training and metadata structures that do not match how people actually search.

Technology alone does not solve these problems.

What is metadata governance?

Metadata governance defines how metadata decisions are made and maintained.

This may include standards, roles, responsibilities, approval processes and procedures for updating controlled vocabularies or taxonomies.

Governance helps prevent metadata from becoming inconsistent over time.

Who should own metadata?

There is no universal answer.

Metadata ownership may involve Digital Asset Management teams, librarians, archivists, marketing operations, content teams, data governance groups or other stakeholders.

What matters is having clearly defined responsibility and decision making authority.

How do you know whether your metadata is working?

Look at how people search for and use content.

Are users finding the correct assets?

Are searches returning too many irrelevant results?

Are people recreating content because they cannot find existing assets?

Are users relying on folders or personal knowledge instead of search?

These behaviors can reveal metadata problems.

Metadata Consulting

Does Tagging.tech offer metadata consulting?

Yes.

Henrik de Gyor provides consulting related to metadata strategy, Digital Asset Management and digital content operations.

Engagements can focus on specific metadata challenges or broader DAM and content management initiatives.

What types of metadata consulting services are available?

Services may include metadata strategy, metadata auditing, metadata implementation, metadata mapping, metadata migration, metadata normalization, schema development and metadata deduplication.

Consulting may also address taxonomy, controlled vocabularies, tagging processes and Digital Asset Management.

What is a metadata audit?

A metadata audit examines how metadata is currently structured and used.

The review may identify inconsistent values, duplicate terms, missing metadata, unnecessary fields and other problems that affect findability.

An audit can provide a clear starting point for metadata improvement.

What is metadata normalization?

Metadata normalization improves consistency.

For example, a collection might contain:

USA

U.S.A.

US

United States

United States of America

Normalization may establish a preferred value and update inconsistent variations.

What is metadata deduplication?

Metadata deduplication identifies duplicate or overlapping metadata values.

For example, “automobile,” “automobiles,” “auto” and “autos” may be creating unnecessary duplication.

Deduplication can help simplify metadata and improve consistency.

What is metadata mapping?

Metadata mapping defines how information from one field, schema or system connects to another.

Metadata mapping is particularly important during system integrations and Digital Asset Management migrations.

Can you help with a DAM migration?

Yes.

Metadata is a critical part of most DAM migrations or consolidations.

Consulting can include reviewing existing metadata, identifying what should be retained, mapping fields, normalizing values and preparing metadata for migration into a new DAM system.

Do you work with organizations that already have a DAM?

Yes.

A DAM does not automatically guarantee good metadata.

Existing DAM programs may need help improving search, simplifying schemas, updating taxonomies, addressing inconsistent metadata or developing better governance.

Can you help an organization that does not have a metadata strategy?

Yes.

Many organizations begin with metadata that developed organically over time.

The first step is understanding the content, users, and business requirements.

A practical metadata strategy can then be developed based on how the organization actually needs to manage and find assets.

Do you work with beginners as well as experienced DAM teams?

Yes.

Metadata problems exist at every level of maturity.

Some organizations need help understanding basic metadata concepts. Others have established DAM programs and need assistance with complex taxonomies, migrations or governance.

The approach should match the organization’s actual needs and maturity.

Tagging.tech Interviews and Archive

What are the Tagging.tech interviews?

Tagging.tech features conversations with people involved in keywording, image recognition, video recognition, computer vision, crowdsourcing and related technologies.

The interviews explore how organizations and technology providers approach the challenge of describing digital content.

Who is interviewed on Tagging.tech?

Guests include professionals working with tagging technologies, keywording services, artificial intelligence, computer vision and digital content.

The goal is to hear directly from people developing and using these approaches.

Are the Tagging.tech interviews still relevant?

Yes.

Technologies continue to change, but many fundamental questions remain the same.

How should content be described?

What can machines identify?

Where are humans still necessary?

How do organizations evaluate accuracy?

How can metadata improve search?

Earlier conversations can provide useful context for understanding how tagging technologies and services have developed.

Where can I find Tagging.tech interviews?

Interviews and related content are available throughout the Tagging.tech website and archive.

Browse the site to explore conversations about human keywording, image recognition, video recognition, artificial intelligence and other tagging technologies.

Are interviews available as a podcast?

Tagging.tech includes audio interviews and conversations focused on tagging, keywording, and technology.

Availability may vary by interview and platform. Visit the individual interview pages for the available listening options.

Are transcripts available?

Transcript availability may vary depending on the interview, due to permissions.

Visit the individual interview page to see the content and resources provided for that conversation.

Getting Started

I am new to metadata. Where should I start?

Start with a simple question:

How do people need to find this content?

Do not begin by creating hundreds of metadata fields.

Understand the users, content, and search requirements first. Then identify the descriptive information needed to support those requirements.

We have thousands of untagged assets. What should we do?

Do not immediately start manually tagging every asset.

First, evaluate the collection.

Determine which assets have business value, how the content needs to be found, and whether automated tagging or professional keywording services could help.

In some cases, not every asset needs the same level of metadata.

Should we use human keywording or automated tagging?

Possibly both.

Automated tagging can process large amounts of content quickly.

Human keywording can provide context, specialized knowledge, and quality control.

The best approach depends on the content, budget, required accuracy, and scale of the collection.

How should we evaluate a tagging technology?

Test the technology using your own content.

Do not rely entirely on demonstrations using carefully selected sample images.

Measure the quality and usefulness of the tags generated for real assets from your organization.

The most important question is not whether the technology can generate tags.

The question is whether those tags help your users find the content they need.

Need Help With Metadata?

If your organization is struggling with metadata, keywording, taxonomy, search, or simply finding your stuff easily,
consulting services are available.

You do not need to add more unnecessary complexity. The goal is to make your digital content easier to organize, find, use, reuse, repurpose, remain actionable and improve what you are telling your audience.