Wikidata Search Skill

Search and retrieve data from Wikidata, the free knowledge base.

API Endpoint

Base URL: https://www.wikidata.org/w/api.php

Core Functions

1. Search Items (wbsearchentities)

Search for entities by label or alias.

curl 'https://www.wikidata.org/w/api.php?action=wbsearchentities&search=QUERY&language=en&format=json&type=item&limit=10'

Parameters:

• search: Search term (required)
• language: Language code (default: en)
• type: item (Q-entities) or property (P-entities)
• limit: Max results (1-50, default: 7)
• continue: Offset for pagination

Response fields per result:

• id: Entity ID (e.g., Q42)
• label: Primary label
• description: Short description
• aliases: Alternative names
• url: Wikidata page URL

2. Get Entity Details (wbgetentities)

Retrieve full entity data including claims/identifiers.

curl 'https://www.wikidata.org/w/api.php?action=wbgetentities&ids=Q42&format=json&props=labels|descriptions|aliases|claims'

Parameters:

• ids: Pipe-separated entity IDs (max 50)
• props: labels|descriptions|aliases|claims|sitelinks|info
• languages: Filter languages (e.g., en|fr|de)

3. Get Claims Only (wbgetclaims)

Retrieve claims for specific entity/property.

curl 'https://www.wikidata.org/w/api.php?action=wbgetclaims&entity=Q42&property=P31&format=json'

Extracting External Identifiers

External identifiers are stored as claims with datatype external-id. Common identifier properties:

To extract identifiers from wbgetentities response:

# claims = response['entities']['Q42']['claims']
# For each property P:
#   claims[P][0]['mainsnak']['datavalue']['value'] -> identifier string

Python Script Usage

Use scripts/wikidata_api.py for programmatic access:

from scripts.wikidata_api import WikidataAPI

wd = WikidataAPI()

# Search for items
results = wd.search("Albert Einstein", language="en", limit=5)

# Get entity with identifiers
entity = wd.get_entity("Q937", props=["labels", "descriptions", "claims"])

# Get external identifiers only
identifiers = wd.get_identifiers("Q937")
# Returns: {'P214': '75121530', 'P227': '118529579', ...}

Response Handling

Search Response Structure

{
  "searchinfo": {"search": "query"},
  "search": [
    {
      "id": "Q42",
      "label": "Douglas Adams",
      "description": "English writer and humorist",
      "aliases": ["Douglas Noël Adams"],
      "url": "//www.wikidata.org/wiki/Q42"
    }
  ]
}

Entity Response Structure

{
  "entities": {
    "Q42": {
      "type": "item",
      "id": "Q42",
      "labels": {"en": {"language": "en", "value": "Douglas Adams"}},
      "descriptions": {"en": {"language": "en", "value": "..."}},
      "claims": {
        "P31": [...],  // instance of
        "P214": [{"mainsnak": {"datavalue": {"value": "113230702"}}}]  // VIAF
      }
    }
  }
}

Best Practices

1. Rate limiting: Add 500ms-1s delay between requests
2. Batch requests: Use pipe-separated IDs (max 50 per request)
3. Set User-Agent: Include contact info in headers
4. Filter props: Request only needed properties to reduce payload
5. Handle missing data: Not all entities have all properties