New Data Spaces

FACES

FACES (Feasibility, acceptance, and data quality of new multimodal surveys) is part of this SPP. Project website: texttechnologylab.github.io/FACES.

Conventional surveys record the answer and discard everything that produced it - the hesitation, the gaze, whether the question was understood at all. FACES asks whether conducting interviews in VR can retain that context, and whether the resulting data holds up: is a VR-mediated interview feasible, do respondents accept it, and is the data quality comparable to established modes?

The studies conducted in FACES are listed on the FACES website.

InterView

The platform built to answer that question is InterView. A participant wears a VR headset and sits across from the interviewer's avatar in a shared virtual room; the interviewer works from an ordinary web browser. Speech, gaze, posture, hand movement, facial expression and the answers themselves are all captured on a shared clock.

InterView is built on Va.Si.Li-Lab and adds a WebRTC media layer, a browser-based interviewer console, and a token-driven session model. Avatar emotions come from ReEmote. Both are listed under Systems using Va.Si.Li-Lab, their papers under Publications.

InterView has its own documentation

Architecture, the full deployment guide, and reference documentation for every service - Janus, coturn, Ubiq, the logging API and the web client - are at texttechnologylab.github.io/InterView.

Setting up the VR client is documented here, in InterView VR Client.

Repository	Contains
InterView	Deployment stack and platform documentation
Va.Si.Li-Lab	The VR client
faces-website	Web client
Va.Si.Li-Lab-backend	Ubiq server and logging API sources