Knowing that each customer is unique,  we build the models using your data – giving you unparalleled performance.

What is auto-tagging?

The answer to what auto-tagging is goes something like: “Automatically adding metadata to a news article”, but that is not what we talk about when we talk about auto-tagging. That definition leaves you with more questions than it answers. We will help you to understand it more!

What is metadata & why is it important?

Why can’t journalists add it themselves?

What kind of metadata are we talking about?

How do we get it into our system(s)?

Why pick iMatrics?

How does auto-tagging work?

Do you support any taxonomy standards?

What are the technical details?

What is metadata & why is it important?

Metadata is important because it is the foundation of all business intelligence. We consider it our mission to make the media industry data driven, and the foundation of any data driven company is data.

    Use automatic metadata to better index your content.
    • Relates to topic/theme pages.
    • Relates to SEO improvements.
    • Use to easily exchange content and connect different types of media.
    Can be used as a basis for contextual recommendations
    • Improves user experience and keeps the user on your site.
    • Increased page views.
    Use metadata for personalization.
    • Relates to followable tags and push notifications.
    • Enables the creation of “My Newspaper”.
    Gain a deeper understanding of your content.
    • Find production inefficencies.
    • Combine with user data and follow what your users are interested in.
    • Basis for user segmentation.
    Enable site automation.
    • Allow the metadata control the initial placement of your content.
    • Aim your editors’ efforts towards fine-tuning the placement of important news.
    Reimagine productivity tools.
    • Use the metadata as a basis to suggest good images for your content.
    • Or to suggest related content in order to easily create packaged stories.

    Why can’t journalists add metadata themselves?

    There are two very short answers to this question. The first is that they don’t want to. The second is that since they don’t want to, they are bad at it. Mistakes and inconsistencies are simply a part of reality when you have multiple people tagging their own content.

    Our automatic solution is quick, consistent and precise which in turn leads to much higher quality of metadata.

    Four journalists choosing different but similar topics for a news article

    What kind of metadata are we talking about?

    We primarily work with the type of metadata that is commonly referred to as tags and we usually refer to them as concepts. For these concepts we have three main types. These are Categories, Topics and Entities.

    Sometimes we also work with other types of metadata. For example when working with gender analysis, sentiment/emotion, content type or news value. These areas are usually part of a future step and if you are wondering what we do in these areas, please reach out to us!


    Categories are high-level subjects with a specific purpose. This purpose could be to hold up the main site navigation or simplified content/production/interest analysis. Usually, we recommend 1 – 2 categories per article and that they should indicate the main aspect(s) of an article. The category part of the taxonomy usually isn’t changed very often.

    Also, it is possible to integrate standards such as IPTC Media Topics or IAB content taxonomy as part of this.


    Topics are more in-depth and detailed. Maybe you want to have a topic called Tea, for all your Tea interested users. Generally topics can be used for detailed site navigation/automation, SEO including topic/theme pages, personalization functionality such as follow and get push notifications or personal news feeds and detailed user interest analysis.


    Entities are named items of any of five subtypes. These subtypes are Person, Place, Organization, event and Object. We usually have a common and natural understanding of Person, Place and Organization, but Event and Object usually raise some questions. Therefore we will clarify them. We define an Event as limited by time, such as the Summer Olympics 2016. An object could for example be a book. The use cases of entities are similar to that of topics.

    How do we get it into our system?

    One of our strongest offerings are our integrations with our CMS partners. If your CMS vendor supports us it’s as simple as activating a plugin!

    If you like to do things your own way, we are still there for you. You can integrate our service practically anywhere through our API. If you need/want we can also do the integration for you according to our consulting fees.

    Superdesk Logo

    Why pick iMatrics?

    Tailored for you

    • We know that each customer is unique – so we build the models using your data – giving you unparalleled performance for your data.
    • The world is in rapid change and the media industry is no exception. Flexibility is super-important. Either through your own input or via the industry standards you can customize the behavior of your taxonomy. Want to tag sports with every football? – No problem.
    • Via supereasy GUI tools or API calls you can control your data and easily change the taxonomy and influence how it is tagged. Just provide a few keywords and we will sort out the rest.
    • Without any downtime and within minutes your changes will apply to your service allowing you to make fast and rapid changes.

    Infinitely scalable

    • By using our service your business will prosper, so you might need to scale up. Then it is good to know that our service can scale up with you either permanently or temporarily at ease by utilizing AWS.
    • iMatrics tagging service was built with speed in mind so re-tagging your articles while you write, to see their effects, is not an issue.

    Innovation partner

    • Our premium support will guide you through all your metadata challenges.
    • We are available for innovative projects to help you take your system(s) and site(s) to the next level.
    • Be a part of the iMatrics Metadata & Media Innovation hub where we showcase the latest technology and use-cases, arrange customer experience exchange events and much more will be added in the future.

    How does auto-tagging work?

    The technical term for programs that analyze text is natural language processing, and there are thousands of engineers and researchers working in this field. That means there are fantastic resources to leverage. Our system uses a mix of state-of-the-art frameworks, traditional rule-based business intelligence, and our own proprietary AI methods. We combine these powerful methods with comprehensive databases based on open data such as OpenStreetMap and Wikidata to achieve the best result possible.

    In short, we like cherry-picking the best parts of different methods, as well as developing our own.


    Do you support any taxonomy standards?

    Today we support IPTC Media Topics and Wikidata IDs. We are also planning on adding IAB dual taxonomy tagging support.

    What are the technical details?


    Spring Boot

    Open knowledge bases

    We regularly sync open knowledge bases such as Wikidata, Wikipedia and Openstreetmap to provide along with the tags. This means that you can get a reference to these knowledge bases in order to get e.g. ids, descriptions and geodata.
    The open knowledge bases provide you with millions of high quality entities to start with.



    Teknikringen 9, 583 30 Linköping Sweden

    Call Us

    +46 70 746 4224

    We use cookies in order to give you the best possible experience on our website. By continuing to use this site, you agree to our use of cookies.