May 5, 2022 by Eveline de Boer
Smartocto is one of 33 startups to join the REACH incubator programme!
REACH is a European-wide second-generation incubator for data-fuelled start-ups & SMEs that aims to facilitate the development of trusted and secure innovative solutions based on industrial and personal data. 33 startups impressed the jury with their solutions to corporations’ big data challenges and were accepted into the REACH programme!
Smartocto sees itself as a global innovator and a frontrunner in media intelligence, and everything we do is predicated on the belief that data solutions must be transparent, helpful and cooperative - values also shared by those at REACH. We also know that to get really valuable and persuasive use cases, synergy is vital and our mission cannot be accomplished in a vacuum. In order to bring it to the next level, we need support to build and enrich our piloting partners, knowledge, data sets and feedback.
Smartocto’s participation in the REACH incubation programme comes in the form of FLAME: federated learning assisting media entities. Federated learning is a machine learning technique that trains central algorithms without exchanging the data. It sends a structured model to a client, and receives a more trained model back, the learnings and refinements of which can be applied to others. This way, all participants can learn from their peers, without giving up their own individual data.
There are a thousand use cases possible with federated learning, but smartocto selected the press agency ecosystem because we know the need here is a pressing one. Press agencies want to optimise their content strategies, their formats and timing, and they also need to have a sense of the impact, reach, engagement and value that their content is achieving.
The issue? Granular data exchanges with clients are considered sensitive.
With federated learning, both client and press agencies can strive for a better fit, without getting entangled in data exchange discussions. We are proud that PA Media and ANP, the press agencies of the UK and the Netherlands will take part in the project.
We think that introducing the federated learning technique in content intelligence will be a groundbreaking way to achieve more synergy and value in media data, while at the same time avoiding risks with data confidentiality and possible issues with GDPR compliance. Ultimately it will enable cooperation between various organisations in the media ecosystem.
Additionally, this first application supports the vital role that press agencies play in journalism. If they can plan next steps in digitisation, organise an even better market fit, and enhance their business and societal value, journalism will be the better for it.
Smartocto will develop the federated learning platform and perform the first training cycle with the participation of the two aforementioned press agencies, and the huge volume of content they have. All kinds of questions can be addressed with FLAME: user needs, conversions, semantics.
After this first stage, we will look for participants to apply the model to their organisations - and we’re sure that they will profit from the results. And if everybody is satisfied with the outcome, we can then look to apply federated learning to different scenarios: user needs and media organisations, benchmarking the impact of topics, or - in real time - the virality of stories.
It’s exciting and new - and it has the potential to achieve significant change…
* This project has indirectly received funding from the European Union’s Horizon 2020 research and innovation programme under REACH Incubator (Grant Agreement no. 951981).
