Program

Tentative half-day program. Slot allocation depends on the number and type of accepted contributions.

Schedule

Date to be announced, TheWebConf 2027, Dublin. All times are local (IST, UTC+1).

TimeSession

Paper assignments to sessions will be posted after author notification. See Accepted Papers.

Interactive benchmark and evaluation design session

SynWeb is built around discussion rather than a sequence of conventional talks. In the interactive session, participants work in groups on three themes. Each group identifies a tractable evaluation task, suitable data sources, baseline checks, and minimum reporting requirements.

Provenance and verification

What evidence, metadata, and validation mechanisms are needed before synthetic data can be reliably reused in web applications?

Contamination and distribution shift

How should synthetic data be evaluated when it interacts with human-generated content, model-generated content, and downstream training or retrieval pipelines?

Downstream web utility

Which evaluation settings show whether synthetic data is useful for search, recommendation, knowledge extraction, content moderation, and web-agent evaluation?

Community roadmap

The organizers will consolidate the session outcomes into a public SynWeb research and evaluation roadmap, including a preliminary benchmark specification and a minimum reporting checklist. It will be linked from this site after the workshop.