A Symbolic Virtual World for Training Agents in Cross-World Exploration
Sawyerr, William (2026) A Symbolic Virtual World for Training Agents in Cross-World Exploration. In: Advanced Information Systems Engineering Workshops. Lecture Notes in Business Information Processing (586). Springer Cham, Switzerland, pp. 1-12. ISBN 978-3-032-28159-3 (In Press)
- Restricted Access Documents
- Details
- 6951:126966
![[thumbnail of paper_553.pdf]](https://research.uca.ac.uk/style/images/fileicons/application_pdf.png "paper_553.pdf")
Cross-world exploration (decision-making across multiple worlds under entry-gated evidence) is hard because decisive evidence about a candidate world requires costly, uncertain entry. We formalize this as a meta-decision problem and present a Symbolic Virtual World (SVW), in which regime parameters are explicit, queryable and manipulable. The SVW is instantiated with Earth-derived structure and constrained evidence access via an Earth-proxy layer. Initial runs under controlled regime changes confirm stable knowledge-growth and exploration-coverage signals with measurable sensitivity to regime variation.
This author accepted manuscript is deposited under the Publisher's Open Access Policy. If you wish to use this manuscript for commercial purposes please contact the publisher, bookpermissions@springernature.com.
Actions (login required)
 |
Edit View |