Moderator Dashboard

Zenya Boost, a product from a local Dutch company, commissioned us to build a moderation dashboard for monitoring comments on their campaign discussion platform.

We built a Vue.js dashboard with a moderation pipeline that first translates comments to English using Azure Translation Services, then analyzes them with the Azure Content Moderation API to detect harmful content such as hate speech, sexual content, violence, and self-harm.

To further support moderators, we integrated two Hugging Face models: one to detect whether a comment is a question, and another to perform fine-grained sentiment analysis across 26 sentiment categories. We also designed a prioritization algorithm to surface the most urgent comments first, alongside visualizations showing sentiment distribution and moderation statistics over time.