Geography

Geography facts for the 194 independent UN member states - capital, continent/region, land borders, area, currency, and official language(s). Fully offline: all data is bundled as static JSON, no external lookups at runtime. Available in English, Danish, German, French, and Spanish.

• Facts
• Usage
• A utility skill, not an educational one
• Data sourcing and licensing
• Known simplifications
• Install
• Development

Facts

• "what is the capital of France" - handles multi-capital countries (only South Africa in this dataset) by naming all of them.
• "what continent is Kenya in"
• "what countries border Germany" - explicit "doesn't share a border" response for island nations.
• "how big is Brazil" - area in square kilometers.
• "what currency does Japan use"
• "what language do they speak in Egypt" - handles countries with multiple official languages.
• "tell me about Afghanistan" - a combined overview in one sentence (continent, capital, and borders together) - shares its sentence-builder with ovos-skill-geography-practice's teach mode, see DEVELOPMENT.md.

Usage

"what is the capital of France"
"what continent is Kenya in"
"what countries border Germany"
"how big is Brazil"
"what currency does Japan use"
"what language do they speak in Egypt"
"hvad er hovedstaden i Frankrig"          (Danish)
"hvor stort er Brasilien"                 (Danish)
"was ist die hauptstadt von Frankreich"   (German)
"welche währung hat Japan"                (German)
"quelle est la capitale de la France"     (French)
"quelle langue parle-t-on en Égypte"      (French)
"cuál es la capital de Francia"           (Spanish)
"qué moneda usa Japón"                    (Spanish)

A utility skill, not an educational one

This provides FACTS on demand - it doesn't quiz or teach. ovos-skill-geography-practice depends on this package directly (imports its data and lookup functions) for its quiz mode, the same relationship ovos-skill-unit-practice has with ovos-skill-convert.

Fixed intents are the primary path - this package's curated offline dataset should always be authoritative for its own domain, not compete on a confidence score against online knowledge skills (e.g. Wolfram Alpha) for the same question. A narrow Common Query safety net also exists (handle_common_query()) for when the platform's own routing sends a question elsewhere before our intent gets a chance - found via live testing, see DEVELOPMENT.md.

Data sourcing and licensing

Country/capital/region/border/area/currency/language data from mledoze/countries (ODbL-1.0), trimmed to the 194 independent UN member states. Country/ currency/language names in da/de/fr/es come from Unicode CLDR (via Babel + pycountry, used only as one-off data-generation tools, not runtime dependencies). Region/subregion names are hand-translated. Full attribution and licensing details: CREDITS.md, including a note on a similar prior skill (OVOSHatchery/ovos-skill-countries) that doesn't work today (dead API since 2021) and why this is a fresh build rather than a fix to that one.

Known simplifications

• Capital names in da/de/fr/es are partial, best-effort - no CLDR-equivalent source exists for city names. See CREDITS.md.
• Language name coverage is ~86-89% (CLDR + pycountry) - the remainder (obscure regional/minority languages) fall back to the source dataset's own English name rather than a guessed translation.
• Spoken country names with a leading article ("la France", "die Türkei") are handled with a simple prefix-strip (ARTICLE_PREFIXES in __init__.py), not full grammatical parsing.
• No population or timezone data - population figures go stale quickly; timezone isn't in the current source dataset's schema at all (see CREDITS.md).
• Scope is 194 independent UN member states only.

Install

pip install ovos-skill-geography

Development

See DEVELOPMENT.md.

Category

Information

Tags

#geography #reference #capitals #countries