Abstract:
Mask App is a full-stack social networking platform built to test how modern
moderation tools and credibility signals can fit naturally into a familiar social feed.
It offers all the expected features-posting, commenting, reacting, sharing,
bookmarking, and creating groups or pages with scoped visibility-while adding realtime notifications and private messaging secured with AES-GCM encryption. A
server-side link-unfurl pipeline generates rich previews for shared URLs, while an
admin console supports policy enforcement and content removal across user, page,
and group contexts. A distinguishing element of Mask App is its built-in fact-check
and trust scoring workflow. Text posts may be evaluated by an external model (Google
Gemini) through a backend API. Checks are triggered by engagement thresholds or
an optional author tag and are guarded by rate budgets to control cost and load. The
model returns a structured verdict with an explanation and confidence, which is
surfaced as a compact “context pill” on the post. Separately, the system aggregates
historical outcomes into an author-level trust score and tier. This score drives a
visible Trust Badge and enables ranking adjustments in discovery views, offering
users lightweight, interpretable signals without hiding content outright.
Architecturally, the frontend is a React application, and the backend is a
Node/Express API backed by MongoDB, with optional Cloudinary storage. The
project is deployable on common student-friendly platforms (e.g., Vercel for the
frontend and Render for the backend) using environment-driven configuration.
Overall, Mask App demonstrates a practical approach to credibility features in social
media while preserving usability. Future work includes broader multilingual checks,
richer analytics for admins, and expanded privacy controls for messaging.
