How the SAV public demo is produced
The demo shows SAV output from a fixed seven-paper corpus. It demonstrates the research workflow; it does not replace the papers or provide citation-ready conclusions.
Who maintains it
SAV maintains the product, demo corpus, and published snapshot. Questions or corrections can be sent to sav@yz.tr.
How papers are processed
PDF text and document structure are extracted first. Language models then produce draft briefs and candidate arguments from that extracted text. Evidence passages remain attached to generated claims so readers can compare each claim with its source context.
What is automated
Briefs, argument labels, summaries, and theme synthesis may be AI-generated. Their wording can contain errors, omissions, or interpretations unsupported by the cited passage. The original paper remains authoritative.
How to verify output
Open the source paper, inspect the linked passage and surrounding context, and decide whether it supports the generated claim. Check page numbers and extraction quality against the PDF.
Source ownership
Paper titles, authorship, quotations, and linked publications belong to their respective authors and publishers. Cite the source paper, not SAV output.