Meta-Tagging Platform

עברית
A tag layer for academic literature in which every tag carries the exact sentence from the paper it was read from.
🔍

Find the definition that fits a corpus

live
enter
Choose which papers make up the corpus and see immediately which definition of the concept fits it best — and exactly which cases it fails on.
  • Free choice of papers
  • Ranking updates instantly
  • Under every score, the case table and the quote
📐

The method, and its limits

live
read
Not an About page. What the tool does not know, and how to reproduce every number it shows.
  • All data open to download
  • The limits are shown to everyone
🏷️

Tag a paper

manual service
Send a DOI, source link or MHTML. We check the full text, licence and text quality before asking for payment; accepted papers are tagged with exact source sentences and reviewed by Shir.
  • Source check before payment
  • Scans, PDF-only copies and uncleared sources are refused
  • Source check before any payment; full refund if we cannot tag it
⚖️

Score a definition

manual service
Name a term this corpus does not yet define, and we build its board: the rival definitions, the adjudicated cases behind each one, an MCC table and the deliberately circular control that calibrates it.
  • For terms with no board yet — the existing boards are free
  • Every case carries the exact sentence it was read from
  • $2.50; nothing is charged before the term is accepted
the corpus · search · transfers
CC BY 4.0 · source
© Shir Sivroni

What this is

A tag layer for academic literature. Every tag carries the exact sentence from the paper it was read from, so any claim it makes can be checked against the source.

What makes it different

Definitions are not asserted, they are scored. Rival definitions of a concept are ranked against adjudicated cases, and a deliberately circular definition is kept in the table as a control: if it wins, the ranking has measured nothing.

What is not finished

Three concepts have a scored board. The corpus holds far more terms that are defined more than one way, and those are shown without a score. Planned next: the negative space — terms the literature should have coined and did not.

The data is open

Every number here is reproducible from the data files; reference_scorer.py recomputes the scores from them. Also README, llms.txt and other projects.

The tag layer is CC BY 4.0. Paper text belongs to its authors and is linked, never reproduced.