# Question-bank source notices

## Arts expansion (`arts3.json`)

The 27,200 English multiple-choice records in this shard were assembled offline from factual catalogues and lexical entries. The question forms, option selection and spelling distractors are transformations made for this project. The saved bank requires no external service or model at runtime. These sources do not endorse this project.

### The Museum of Modern Art (MoMA)

5,500 Art questions use The Museum of Modern Art Collection dataset, obtained on 16 October 2026 from the museum's official repository: [MuseumofModernArt/collection](https://github.com/MuseumofModernArt/collection). Inputs were `Artworks.json` and `Artists.json`. Each question links to the corresponding MoMA work page.

The museum publishes these metadata datasets under [CC0 1.0](https://github.com/MuseumofModernArt/collection/blob/main/LICENSE.md). Only records marked curator approved (`Cataloged: Y`) were eligible here. Question wording, medium-family grouping, difficulty labels and distractors were added by this project. This is a modified derivative, not an official MoMA question collection. The metadata source includes no artwork-image rights; no artwork images were copied for these questions. Catalogue data is supplied as is and may be revised by its publisher.

### Wikidata and DBpedia

16,200 Literature, Cinema and Music questions use fully received Wikidata credit statements and, where available, publication dates, work types and creator context. Property relations include author (`P50`), director (`P57`), performer (`P175`) and publication date (`P577`). Each question links to its Wikidata item. [Wikidata structured data is released under CC0](https://www.wikidata.org/wiki/Wikidata:Licensing); this bank does not reproduce Wikidata discussion-page prose.

The global SPARQL response streams timed out and were incomplete. Only complete individual binding objects were used; truncated rows were discarded. English titles were checked for uniqueness among all received records, and works with multiple credited creators in those records were excluded. This does not establish global title uniqueness outside the received catalogue. Anonymous and selected traditional attributions were excluded. Dates were never invented.

Additional release-year context for a small subset comes from the public [DBpedia SPARQL endpoint](https://dbpedia.org/sparql), joining work identifiers with Wikidata identifiers. Only complete received date triples were used. These factual dates are context; the main credit source remains the linked Wikidata item. DBpedia is a community extraction of Wikipedia data, and its records may be incomplete or revised.

### Princeton WordNet 3.1

5,500 Language questions use Princeton University's [WordNet](https://wordnet.princeton.edu/), downloaded as [`wn3.1.dict.tar.gz`](https://wordnetcode.princeton.edu/wn3.1.dict.tar.gz) on 16 October 2026. Word meanings and definitions are derived from WordNet 3.1; spelling questions, misspellings and choices were added by this project. The stored fact key retains the WordNet synset identifier. Source corpus frequency is used only as an estimate of familiarity, not as measured player difficulty.

The complete notice below is transcribed from the headers of the downloaded WordNet 3.1 dictionary data files; retain it with copies and derived distributions of this material.

```text
This software and database is being provided to you, the LICENSEE, by
Princeton University under the following license. By obtaining, using
and/or copying this software and database, you agree that you have
read, understood, and will comply with these terms and conditions.:

Permission to use, copy, modify and distribute this software and
database and its documentation for any purpose and without fee or
royalty is hereby granted, provided that you agree to comply with
the following copyright notice and statements, including the disclaimer,
and that the same appear on ALL copies of the software, database and
documentation, including modifications that you make for internal
use or for distribution.

WordNet 3.1 Copyright 2011 by Princeton University. All rights reserved.

THIS SOFTWARE AND DATABASE IS PROVIDED "AS IS" AND PRINCETON
UNIVERSITY MAKES NO REPRESENTATIONS OR WARRANTIES, EXPRESS OR
IMPLIED. BY WAY OF EXAMPLE, BUT NOT LIMITATION, PRINCETON
UNIVERSITY MAKES NO REPRESENTATIONS OR WARRANTIES OF MERCHANT-
ABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OR THAT THE USE
OF THE LICENSED SOFTWARE, DATABASE OR DOCUMENTATION WILL NOT
INFRINGE ANY THIRD PARTY PATENTS, COPYRIGHTS, TRADEMARKS OR
OTHER RIGHTS.

The name of Princeton University or Princeton may not be used in
advertising or publicity pertaining to distribution of the software
and/or database. Title to copyright in this software, database and
any associated documentation shall at all times remain with
Princeton University and LICENSEE agrees to preserve same.
```

### Review limits

Automated checks cover exact category/level totals, normalized prompt and fact-key uniqueness against the preserved 10,000-question baseline, four distinct options, answer membership and literal answer leakage. Title/creator filters and lexical synonym/ancestor filters reduce particular ambiguity risks. They are not an exhaustive human factual or semantic review of every question. Difficulty estimates have not been calibrated through player testing. Original source caches and authoring validation reports are editorial inputs under `.qa/arts3-*`; runtime uses the saved JSON shard.

## Nature expansion (`nature3.json`)

The 5,400 records use accepted species and classification fields from the [GBIF Backbone Taxonomy](https://www.gbif.org/dataset/d7dddbf4-2cf0-4f39-9b2a-bb099caae36c), retrieved through the public GBIF species API on 16 October 2026. Citation: GBIF Secretariat (2023). GBIF Backbone Taxonomy. Checklist dataset [doi:10.15468/39omei](https://doi.org/10.15468/39omei), accessed via GBIF.org on 2026-10-16. The [dataset metadata](https://api.gbif.org/v1/dataset/d7dddbf4-2cf0-4f39-9b2a-bb099caae36c) declares [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).

The retained input is `data/nature3-gbif.json`. Scientific names, available English common names, and family/order/class fields are source metadata. Question wording, broad-group mapping, options, difficulty and selection were added by this project; GBIF does not endorse the questions. Each question links to its GBIF species record. Classification reflects the saved backbone snapshot and may differ from another taxonomic treatment or a future revision. The three levels use broad groups, orders and families respectively; these estimates have not been calibrated with players.

## Food expansion (`food3.json`)

The 5,600 records use the official USDA FoodData Central [SR Legacy April 2018 CSV archive](https://fdc.nal.usda.gov/fdc-datasets/FoodData_Central_sr_legacy_food_csv_2018-04.zip). The compact retained input is `data/food3-usda.json`, derived from food, nutrient, food-nutrient and food-category tables. See the USDA [data documentation](https://fdc.nal.usda.gov/data-documentation/). Individual-source questions link to the corresponding FoodData Central identifier; questions combining several records link to the source archive and retain the involved identifiers in their explanations.

USDA identifies FoodData Central data as public-domain material released under [CC0 1.0](https://fdc.nal.usda.gov/api-guide/#bkmk-9). Attribution: U.S. Department of Agriculture, Agricultural Research Service. FoodData Central, 2019. This project uses the SR Legacy release finalized in April 2018.

Preparation states and units are preserved. Reported amounts are per 100 g edible portion; energy-normalized comparisons calculate amounts per 100 kcal from that entry's reported energy. Portion and mixture questions are project-authored arithmetic scenarios using those saved concentrations, with assumptions and rounding stated. Independent decimal calculations checked every portion and mixture answer; all comparison choices were checked against the reference threshold. The records describe historical reference values, not measured nutrient contents of every current product or personal dietary advice. USDA does not endorse this transformed question bank.

## Mythology expansion (`myth3.json`)

The 5,600 records derive from public [DBpedia SPARQL](https://dbpedia.org/sparql) extracts of Wikipedia-based entries for deities, traditional narrative figures and legendary beings, retrieved on 16 October 2026. These are encyclopedic sources, not an official theological authority. Retained factual relations and the original query URLs are in `data/myth3-relations.json`; retained alias mappings and their query URL are in `data/myth3-redirects.json`. Each question links to its DBpedia resource. Resource pages identify their Wikipedia extraction as [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/); this project's modified `myth3.json` derivative is provided under the same license, with attribution through these notices and record links.

Project changes include English question wording, distractors, difficulty estimates, canonical relationships, redirect resolution, and removal of malformed or alternative statements. Parent/child inverse relations and symmetric sibling/consort pairs are merged into one fact. Pairs having multiple simultaneous family relationships are excluded from questions with mutually exclusive relationship choices. Literal place names are preserved rather than resolved to similarly named people. Specific attributes use alternatives of the same kind, with lexical and selected equivalent-name filters. Harder alternatives favor the same inferred tradition when enough candidates exist; this grouping creates no new factual assertions.

Independent source checks covered all 5,600 recorded answers, and normalized prompts/fact keys were checked against the original 10,000 and these Nature/Food additions. These checks do not establish exhaustive semantic equivalence detection or a human expert review of every source. Accounts vary by region, text and period; explanations refer to the retained account. Difficulty is estimated and requires player calibration. No third-party quiz collection was imported, and no model or external service generates these questions during play.

## World expansion (`world3.json`)

The 21,400 Geography, History, Sport and Culture questions use factual catalogue records from [GeoNames](https://www.geonames.org/), the [UNESCO World Heritage Centre](https://whc.unesco.org/en/list/), [UNESCO Intangible Cultural Heritage](https://ich.unesco.org/en/lists), historical [Olympic results](https://olympics.com/en/olympic-games/olympic-results), and [DBpedia](https://dbpedia.org/). Each stored question retains its source link. These publishers do not endorse the transformed questions. GeoNames data is used with attribution under its [CC BY license](https://www.geonames.org/export/). The DBpedia-derived material in this modified shard is distributed under CC BY-SA 4.0, with attribution through this notice and each resource link. Dates and locations reflect the retained source records. No copyrighted sports or heritage descriptions were copied as question prose.

## Space and calculation expansions

`space3.json` uses a saved [NASA Exoplanet Archive Planetary Systems Composite Parameters](https://exoplanetarchive.ipac.caltech.edu/docs/API_PSCompPars.html) snapshot; attribution: NASA Exoplanet Archive, [doi:10.26133/NEA13](https://doi.org/10.26133/NEA13), retrieved 16 October 2026. Question wording, numeric models, assumptions, rounding and alternatives were added by this project. Published estimates are identified as reference values; hypothetical calculations do not assert measured surface conditions.

`math3.json`, `science3.json`, `business3.json` and `society3.json` contain original numerical scenarios and question families. Linked [OpenStax](https://openstax.org/subjects) references describe the underlying subjects; their textbook prose and exercises were not copied. Scenario parameters, assumptions and rounding are part of the questions. Difficulty is an editorial estimate. `technology3.json` contains original JavaScript and SQLite exercises; outputs were evaluated offline against a JavaScript runtime and SQLite, with the [ECMAScript 2024 specification](https://262.ecma-international.org/15.0/) and [SQLite documentation](https://sqlite.org/docs.html) retained as references. No user code is executed during play.