Loading the generated data…
This view is withheld.
Work similarity: where each work sits between the two ends of this archive
- Early references
- Recent references
- Candidates
Click a mark to open it. The plot is focusable: press Tab to reach it, then ← → to walk works in publication order, Page Up/Page Down to jump fifty, Home/End for the ends. The complete table below carries every work, including the ones that place no mark.
Selected work
No work selected.
Top term contributions
| Term | Count | Share | log ratio | Contribution |
|---|
Every work
| Title | Year | Group | Scored against | Margin | Prediction | Correct | Tokens | Retained | Retained share | Work id |
|---|
What the two reference groups look like in aggregate
Characteristic words, from pooled counts
Most recent-leaning
Most early-leaning
Vocabulary floor: why 20
| Min doc frequency | Retained types | Retained share | Scored works | LOO accuracy | ROC AUC | JS (bits) | Sign agreement | Margin correlation | Selected |
|---|
What this does and does not say
Refusals
Pattern frequency: one frozen query across every year
This is a frozen, AQL-derived snapshot.
- One or more works at that exact coordinate
- Exact zero: the query selected nothing in that work
Marks are never jittered, so works sharing a coordinate exactly are drawn as one larger mark. Nothing is interpolated between sampled years. The plot is focusable: ← → walk works in publication order.
Selected work
No work selected.
Every work
| Title | Year | Occurrences | Authorial tokens | Rate per million | Exact zero | Work id |
|---|
Totals
Pooled by decade
A decade row pools that decade's numerators and denominators. It does not interpolate, and it does not stand in for an unsampled year.
| Decade | Works | Sampled years | Works with ≥1 | Exact zeroes | Occurrences | Tokens | Pooled rate | Median work rate |
|---|
Pooled by sampled year
| Year | Works | Works with ≥1 | Exact zeroes | Occurrences | Tokens | Pooled rate |
|---|