Work in progress AI integration for the participation platform that allows platform admins to install and configure a spam detection service so that any suspicious content gets reported at the earliest possible step.

"This service can be trained either using the datasets that we provide, or using your own content. This way the engine learns your own business domain and it is able to detect spam or things that are not relevant for your activity." There is also functionality to allow real-time training on instance-specific content, which is informed by admins marking specific content as spam.

The service uses a Bayesian strategy. Admins can also integrate other spam detection services.

Like the rest of Decidim, it can be applied to specific platform modules. So if not all modules are receiving spam, admins don't need to apply the AI spam filter to them.