Abstract
The VIS-3 project developed and tested an Intelligent Virtual Assistant (IVA) for social science data collection, with a focus on housing research. The project combined applied AI development, human-computer interaction, and survey methodology to explore whether a virtual assistant could support respondents in completing an interactive questionnaire with visual content.
A prototype IVA was custom-built by integrating several technical components, including a browser-based survey interface, speech-to-text, text-to-speech, a conversation server, a decision service based on a Large Language Model, avatar modelling, and a response database. The assistant enabled respondents to choose an avatar, listen to spoken questions, read captions, view images, and answer either by voice or by clicking.
The system was tested during HomeEXPO 2025 in Luxembourg. In total, 194 responses were collected, including 103 surveys administered by the Virtual Assistant and 91 self-administered surveys. The questionnaire addressed housing preferences through visual examples related to tiny houses, collective parking, and green spaces in residential neighbourhoods. Respondents also evaluated the images, the survey experience, and the Virtual Assistant.
A prototype IVA was custom-built by integrating several technical components, including a browser-based survey interface, speech-to-text, text-to-speech, a conversation server, a decision service based on a Large Language Model, avatar modelling, and a response database. The assistant enabled respondents to choose an avatar, listen to spoken questions, read captions, view images, and answer either by voice or by clicking.
The system was tested during HomeEXPO 2025 in Luxembourg. In total, 194 responses were collected, including 103 surveys administered by the Virtual Assistant and 91 self-administered surveys. The questionnaire addressed housing preferences through visual examples related to tiny houses, collective parking, and green spaces in residential neighbourhoods. Respondents also evaluated the images, the survey experience, and the Virtual Assistant.
| Original language | English |
|---|---|
| Type | Seed Grant Presentation (ccEXPAR seminar) |
| Media of output | Presentation |
| Number of pages | 36 |
| Publication status | Published - 19 May 2026 |
Keywords
- virtual intelligent assistant
- AI
- multimedia surveys
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver