Periodic Reporting for period 2 - CPN (Content Personalisation Network)
Période du rapport: 2018-09-01 au 2020-04-30
The objective of CPN is to provide a platform delivering a set of services allowing both large and small media companies to target the delivery of news in the right format at the right time to the media consumer.
In Year 1, the vision of the project was defined. Following this vision, several scenarios & use cases have been defined. These use cases then allowed the consortium to define the technical & user requirements for setting up the recommender and the technical platform. In Year 2, the consortium worked hard on improving the platform infrastructure and developing the technology bricks that had been defined. These features were carefully implemented and linked to the recommender framework.
During the final year of the project, focus has been put on improving the recommendation engine as well as the front-end modules. This was done by means of online & offline testing.
These technical improvements were then extensively tested during a one-month pilot with a big group of users. The set-up, evaluation methodology & recruitment for this pilot was carefully addressed. An open pilot was conducted, meaning that there was no fixed start and end date for the testers, with almost 5000 testers. These readers were able to read their personalised article stream via the CPN app (DW & DIAS content), via the DIAS website or via the VRT MYNWS app.
The user satisfaction has been measured by means of click data analysis & regular surveys, which lead to interesting conclusions.
Finally, the consortium also focused on the dissemination and exploitation of the services developed along the project. As organizing a final live event to share the project results was no longer possible due to Covid-19 regulations, a webinar called “Trust, Transparency and Personalisation – Building deeper relationships with your readers” was organized in May.
Based on state-of-the-art studies on system architecture and a testing phase, the microservices architecture was selected. A first version of the CPN Virtual Open Platform was implemented. Following the guidelines from the reference system architecture, all technical partners deployed the first version of their Technology Bricks within the CPN platform.
-Recommender, released by LiveTech
-User Modelling, released by LiveTech
-Personal Data Receipts released by DigiCat
-Relation Extractor released by Imec
-Producer's App released by ENG
All the services offered by the platform were exposed through the API Gateway for end user applications. To demonstrate and to pilot the offered functionalities of the platform an existing application from ATC was adapted to serve as a front-end to show the personalised news stream. Next to this wireframes were sketched-out and validated condensing functionalities requested. These wireframes will be converted to a mobile application to serve as a front-end testbed in Y2.
In regards to GDPR compliance, a version of the so-called Personal Data Receipts (PDR) were developed that will be used to comply with this regulation during the initial piloting in Y2.
In the first year different communication channels were set up (website, social media). A communication plan was established and the first communication materials were produced and published. During this period the newsletter was published along 8 different blog posts. The project was also represented on 10 different events.
In regards to exploitation we focused on updating the market study, exploitation plans and the IPR strategy.
The expected result for the end of the project is a platform for news publishers, for whom the services like personalisation can be integrated in their own infrastructure or who use the platform in conjunction with a reader app, so smaller news agencies can team up and deliver a joint personalised news experience. The platform is built using privacy-by design rules, giving news consumers full transparency and control over their data. The functionality provided will enable better personalisation and insights in usage patterns, allowing all media companies to better serve their audience.