Wikiparse / public prototype

Make the shape of a corpus inspectable.

Wikiparse parses encyclopedic dumps into citable, analyzable form. The parser is published on PyPI (v1.0.2), the first corpus epoch's metadata tables are on the Hugging Face Hub, and this page shows what receipted corpus data looks like when it is made inspectable.

Two data seams, labeled. The thinkers timeline below renders the thinkers-v1 receipted selection from epoch enwiki-20260601. The category explorer stays a hypothetical, versioned sample while the full corpus release clears its evidence gates.

Explore the thinkers timelineDataset on Hugging Facewikiparse on PyPI

Campaign frontier

Five connected pieces of work

  1. 01

    Publish useful subsets

    Small, honestly scoped releases can be more useful than waiting for a single total release.

  2. 02

    Partition for analysis

    Categories should make useful properties cheap to filter, compare, and navigate.

  3. 03

    Separate disambiguation pages

    Keep their relational value while avoiding treating them as ordinary articles.

  4. 04

    Raise citation correctness

    Identify authors, titles, and publishers with evidence rather than superficial success.

  5. 05

    Tell deeper analysis stories

    Use transparent visual analysis for citations and the category tree when the evidence is ready.

What this page proves

A static explanation first; replaceable data seams second.

The page does not read the canonical store, a warehouse, private paths, or an API. Its client enhancements fetch two public JSON files, validate every payload against a closed contract, and leave the reading path intact if any request fails. Timeline numbers trace to receipts on the immutable epoch; explorer numbers remain invented examples.

Thinkers over the ages · receipted corpus data

A timeline of systematic thinkers.

Wikipedia enwiki-20260601 · thinkers-v1 receipted selection

Every point is a person article in the enwiki-20260601 corpus selected by the receipted thinkers-v1 stage: rows are chosen by spine category count with a three-fold weight for philosophy, cognitive science, and the systematic sciences. Years come from Wikipedia's year-of-birth and year-of-death categories; century-rounded rows are marked. Selected rows without a birth year stay in the dataset and its counts, but cannot sit on a birth-year axis. Nothing here is hypothetical: the dataset links its receipts.

Loading receipted data…

Years −800 – 2020

Thinkers in the current view

The table fills when the receipted data loads. It lists matched thinkers in the order the canvas draws them.

Thinkers matching the current filters and year window
ThinkerYearsFieldCategoriesDetails
Awaiting receipted data.

Illustrative explorer

Follow a small category and citation sample

Hypothetical data. The browser loads a versioned, invented sample after this page is displayed. It is a publishing demonstration, not a finding about Wikipedia.

Example data will load when JavaScript starts. The explanatory page remains useful without it.

Filter the illustrative articles by text and category.

Articles
Categories
Citations
Publishers

Category ranking

  1. Awaiting hypothetical data.

Publisher ranking

  1. Awaiting hypothetical data.

Results and graph relations

Articles in the current selection

Hypothetical articles selected by the filters
ArticleCategoriesCitationsScore
Example data is loading.

Graph equivalent

Relationships in the current graph

All selected hypothetical graph relationships
SourceRelationTarget
Example graph relations are loading.
Bounded relation graph. The relationship table above is its complete semantic equivalent, including selected articles, categories, and reachable category-parent relations.