Press agencies are the first to collect news, spread it to media outlets and thus make news scalable. But in the last decade this service has gradually come under pressure. In the digital age it is extremely important to know how your content performs in order to optimise it, but press agencies have no owned platform and thus no access to readers’ data.
With the FLAME solution, federated learning enables one to train an algorithm or model across multiple media stakeholders’ data, without actually exchanging the data. Smartocto will create a powerful impact model of content and a simple interface, and will send it to multiple media outlets that can easily train the model based on their content and performance data. A segment of the model – on press agency content – can be used to inform and notify agency organisations like PAmedia. At the same time, the exchange of data from one partner to another is not necessary, only trained models (increments) are exchanged.