Rhetoric: a new tool for comment moderation

Introduction

Flemish news outlets VRT NWS and Het Nieuwsblad were looking for new moderation tools for their news editors, to mitigate inauthentic behaviour and enable more constructive dialogue, enabling civil discourse by managing textual and visual comments more efficiently.

Challenge

The key to detailed analysis of rhetorical structure and participant identification is stance and role detection, and classification of disagreement types. It is not sufficient to identify word-based individual terms. Image analysis is quite the challenge as well. Memes have more interaction and spread the message faster. They often contain both text and a background image. Many existing tools enable discussion, but do not empower and support citizens in their online debate. Neither do they stimulate citizens to engage in civil and thoughtful discussion.

Solution

We developed new state-of-the-art tools to detect polarising topics and multimodal stance by extracting argument structure. A new role detection engine can identify a participant’s role, and a new classifier can measure the type of disagreement. Convolutional Neural Networks (CNNs) make stance detection happen, and to interpret images we will map the feature space information from the images with the textual content. The machine learning pipeline has built-in features that explain and illustrate the result of the AI and why it chose the result it generated ((Explainable AI). All this comes in tailor-made dashboards for news editors, journalists, moderators and conversation managers, offering metrics and visual analytics. They will be able to check the current state of stance and role detection, along with the classification of disagreement types in a specific conversation.

Result

More information: Rhetoric website