New content management system relies on AI language model
16 July 2024 – The DFL Deutsche Fußball Liga applies innovative technologies to supply national and international Bundesliga and Bundesliga 2 fans with comprehensive content. For example, its new AI-based content management system enables the DFL to tailor the content on its platforms precisely to each user group while providing more choices. The DFL developed this new system to better meet the requirements of the Bundesliga website and the related app than the off-the-shelf system that was used before.
“We wanted to have more control over four key areas,” explains Björn Rosenthal, Head of Product at DFL Digital Sports GmbH. “We needed a better interface with the official match data residing on the Sportec Solutions system; and we needed better access to the media archive.” The database of the DFL subsidiary Sportcast is the world’s largest archive of football footage. “We also see major potential in being able to broadcast content in several languages within a narrow time frame and are using Artificial Intelligence (AI) to accomplish this,” Rosenthal adds.
New languages, new requirements
In today’s globalised world the DFL wants to communicate more closely with Bundesliga enthusiasts living in many different cultures and speaking many different languages. Therefore it is essential to offer content in localised form. This will strengthen the bond with fans and extend the international reach of German professional football, says Rosenthal. “Right now we have a German and an English editorial team,” he explains. “To be more present in other markets, we will offer our content in a range of additional languages – such as Brazilian Portuguese, Spanish and Japanese.”
Establishing a dedicated editorial team for each of these languages would cause excessive costs, however. On the other hand, simply translating existing German texts would not be very helpful either because the content itself needs to be adapted to the target audience, apart from the language: for example, fans in Japan are naturally more interested in Japanese players in the Bundesliga and Bundesliga 2, while other topics may be of greater interest to the German-speaking audience than to fans in other countries.
AI language model handles most of the effort
“We realised that we would get better results by using an AI-based language model instead of just translating texts. Today we instruct a generative AI solution to generate an article based on the existing German text, applying specific criteria to optimise it for the target country and language.” The instructions given to the AI solution can include specific data points that are not in focus in the original German text. These might be performance data for a particular player that would put the spotlight on that player in the localised version.
“Today we instruct a generative AI solution to generate an article based on the existing German text, applying specific criteria to optimise it for the target country and language.”
Björn Rosenthal, Head of Product at DFL Digital Sports GmbH
But the text adapted for the target audience is not released to the world unchecked: “Our copy editor reviews the output of the generative AI and optimises, amends and refines the content,” says Rosenthal. “This entire process enables us to extend our international reach very effectively with limited means.”
The DFL implements AI-based technological innovations such as these in close cooperation with Amazon Web Services (AWS) to meet the needs of followers on the Bundesliga digital channels. DFL Digital Sports will continue to refine its AI-based solutions to cater for new markets.