Skip to content
Customer Story

AWS Cloud & Machine Learning Help RASP Focus On Quality Content

Innovative Polish publishing house optimizes its publications for commercial partners with image recognition tool that streamlines editorial process

Related services
Cloud, Amazon Web Services (AWS), Machine Learning

Technology | Telecom | Media


Ringier Axel Springer Polska (RASP) is part of the European media group Ringier Axel Springer Media AG and is one of the largest press publishers in Poland. In its portfolio, the company has over a hundred and seventy titles and websites, among them Onet, Fakt, Forbes, and Newsweek Polska. Utilizing the millions of users across those titles, RASP creates innovative business products that provide commercial partners with a range of high-quality solutions. The organization is highly experimental, allowing its team room to make mistakes in order to creatively and positively impact the digital media industry.

RASP wanted to develop an automated image search tool to give editors time to focus on text content
Using Amazon Rekognition features allowed image tags to be added automatically, ready for search
Editors can find images quickly within new module, and RASP has a platform for future innovation

Manual Image Search: Inefficient Use of Time

As one of the largest media companies in Europe, RASP publishes thousands of online articles daily, relying on an efficient and reliable editorial process. However, the existing tools for image search and upload for articles were starting to slow editors down.

The company came up with the idea to build a cloud-hosted image recognition tool based on machine learning algorithms to help streamline the workflow. By automating the metadata functionalities, the company would make internal library creation and image tagging simpler, faster and more effective. Editors could then focus on writing and refining text content rather than spending time manually searching for relevant images.

A Machine Learning Prototype

RASP worked with Xebia to create a prototype for the project based in the AWS Cloud, using AWS Lambda. Within images, Amazon Rekognition was used to recognize people as well as thousands of objects and scenes, including accurate facial analysis of information such as gender, age and emotions — all vital metadata for the project. The machine learning algorithms used could also recognize logos and commercial signs of specific brands 

Using these tools within the prototype meant that attributes and features would be automatically added to images in the form of tags, making it much easier for editors to select and categorize the most appropriate images for their content. The solution also allows for rapid model development thanks to deep learning algorithms, enabling RASP to easily innovate in the future.

“Thanks to Xebia’s Scrum approach, reliable documentation, high commitment and great communication, the project was delivered on time and to our high expectations. The efficient implementation of the prototype has helped us confirm that we need to roll out the module across our editorial systems.”
ringier axel springer
Katarzyna Ludka Artificial Intelligence Director, RASP

Future-Proofed Image Search

Thanks to the image recognition prototype, RASP demonstrated that integrating the module within its editorial system would be effective in meeting the company’s goal of saving valuable time within the editorial process. In just four weeks, the prototype was built to recognize thousands of objects within images and accurately analyze detailed image attributes. It can now automatically create Polish-language meta-tags for the company’s image content — previously a manual task.

The image recognition prototype was also able to highlight to the company just how quickly even the most technologically advanced applications can be developed on the AWS Cloud. The rapid model development capabilities built into the solution ensures that RASP can continue to creatively innovate into the future.

Explore related customer stories