From a traditional press agency to an innovative thought leader

Metadata, AI, Smart iMage suggestions, auto-tagging, Wikipedia, Wikidata

NTB's journey from a traditional press agency to an innovative thought leader

At the heart of NTB's digital transformation lay a key element - reliable metadata. This isn't just data; it's information about information. Categories, detailed topics, people, places, organizations, events – all these became crucial in tailoring content to meet the precise needs of publishers and, ultimately, their readers.

In this story, we'll dive into how NTB leveraged data to set new standards in the digital news distribution.

For decades the Norwegian Press Agency has been supporting the Norwegian media landscape with hundreds of news articles every day. The articles from NTB provides the client's readers a broad variety of topics, locations and happenings around the world. To make articles from NTB match the demand from both clients and readers. 

These articles complement NTB's clients ordinary news feed, with high quality content, covering a broad variety of topics, locations and happenings around the world.


While the competition from tech giants such as Google, Spotify and Apple are growing their interest in news content; NTB and their clients needed to digitalize. Which means for the clients within publishing offer their readers a more relevant and personalized experienced. 


An important corner-stone in that is reliable data. And for publishers that means metadata (information about information, for example: categories, topics, people, places, organizations, events etc).


"News articles without good metadata is like finding a needle in a haystack" - Tor Kristian Flage

Metadata on articles automate the process of distributing content, to the right people, at the right time, about what they want to read about, and could be interested to read about.


Having that in place, you will offer your readers a great experience. They will read more news, return more often to the site. They will easier find the content they are looking for, and get recommended. Here is where NTB wanted to help. 

However, getting metadata is both time-consuming, and complex. For humans.

Why is metadata a challenge for publishers? (according to themselves)

Time

Tagging articles manually takes around 7 minutes per article for the editorial staff.

Complexity

Same things can mean different things depending on how it is described, or the focus of the article (just take Covid, Corona, pandemic, virus, Covid-19 for an example). And the tags must fit the way the news organization are writing - also called taxonomy.

Consistency

Which makes it impossible for people, even with training. The task might look simple enough, but it ain't.  As Tor Kristian describes it, “having thousands of articles without good metadata is like looking for a needle in a haystack.”

Tor Kristian Flage and Kåre Henriksen, NTB

Partner up with iMatrics


NTB has a strong focus on utilizing technology, that makes things easier both for their editors and their clients. When it comes to metadata NTB, wasn't only looking for a tech solution that delivered an out of the box solution, but rather an partner that could join in NTB's vision. Luckily for this story, the Swedish AI company iMatrics matched NTB's demands.

As Tor Kristian puts it: “We are good at creating content, not at tagging news articles.”.


After an onboarding set up iMatrics auto-tagging SaaS system to fit NTB's taxonomy, and implement it into NTB's CMS system Superdesk. NTB's writers found themself with an integrated tool providing them with metadata tags automatically.

Innovation and collaboration

The demand for innovation and being a press agency in forefront of technology. The next project Wikidata Live Update, was born. The reason behind it was NTB's vision of having one ID on their tags that is the same through all their organization. 


Enabling more and better data for analysis, searching and knowledge graphs. The project we at iMatrics choosed to call Wikidata Live Update was born. And from now on all of our customers has the databases of Wikidata updating the data on their entities, as well as adding new ones in a structured way. Live in their system.


Becoming a trailblazer withing this are, we have had the privilege to present this way of working together with Tor Kristian from NTB on several occasions. For example for the organizations EBU and IPTC.


What's up next?

iMatrics' Smart iMage Suggestions tool simplifies the process of finding suitable images for your content. All based on quality metadata.

Matching data from articles with NTB's image database


So what is the next step you might ask?

If you ask Tor Kristian it will be the Smart iMage suggestion tool that will provide image suggestions on articles. Live while the journalist is writing. This will help the editorial staff saves a lot of time, getting good suggestions for the articles. And at the same time prevent the same image being used over and over again.


The Smart iMage suggestion tool is live since a couple of weeks back. And we can't wait getting you more insights about the value NTB and their clients get from it.

Tor Kristian Flage (NTB) and Fredrik Lundberg (iMatrics)

just before the Minds International Conference in Oslo 2021.