Funding granted for imec.icon project RHETORiC

RHETORiC – Reducing Hate with Editorial Tools for Online Reactions and Comments

The internet is an important space for democratic debate, however, social media have become increasingly toxic and polarized. This makes it  difficult for media channels to manage online comments. Most content publishers in Belgium even disabled reader comments on their channels. Furthermore, because social media allows to share memes and visual commentary, it became increasingly difficult to automate comment and content moderation to provide an open and respectful online debate. Numerous attempts by news channels to encourage thoughtful dialogue were mostly ineffective.  The RHETORiC project will therefore design interface elements and machine learning approaches to create semi-automated tools for news editors and media audiences to identify and counter polarization and contribute constuctively to online debates.

RHETORiC incorporates in-depth research along the following directions:

  • Design of interface elements that present cognitive biases and offer templates for constructive argumentation
  • Enable effective semi-automatic comment & content moderation tools both for text and images
  • Create participatory news formats that stimulate online audience interaction

With this research focus in mind, RHETHORiC will develop a demonstrator composed of a moderation dashboard for media channels and commenting tools for social media users. These tools will be connected within the editorial flow of VRT NWS and Het Nieuwsblad.

RHETORiC will provide valuable and novel insights in how to effectively promote civil participation in online debates. The research aspects within this project will be validated through user testing in order to measure the improvement on the experience, equanimity and balance of the debate, this both quantitatively and qualitatively.

Our partners within the RHETORiC project are: VRT, Textgain, Wieni, Mediahuis, Tree Company, and fellow research group KULeuven – Mintlab