kempersc 3c80de87e0 add isil entries

2025-11-19 23:25:22 +01:00

7.1 KiB

Raw Blame History

R-Class (RESEARCH_CENTER) Initial Extraction Analysis

Date: 2025-11-12

Overview

Extracted Wikidata subclass hierarchies for RESEARCH_CENTER (R-class) from base classes Q31855 (research institute) and Q7310853 (documentation center).

Extraction Statistics

Quantitative Results

Unique Classes: 592
Total Terms (labels + altLabels): 3,724
Languages Covered: 119
Base Classes Queried: 2
- Q31855: research institute
- Q7310853: documentation center

Language Distribution (Top 10)

English (en): 1,055 terms
Spanish (es): 251 terms
Japanese (ja): 231 terms
German (de): 220 terms
French (fr): 199 terms
Slovenian (sl): 132 terms
Chinese (zh): 118 terms
Russian (ru): 110 terms
Italian (it): 101 terms
Korean (ko): 79 terms

Key Research Center Types

Major Categories Identified

Research institutes (general)
Think tanks / policy research centers
Documentation centers
Laboratory types:
- University laboratories
- Research laboratories
- Medical research labs
- Biological stations
- Field stations
Academic research units:
- University institutes
- Research departments
- Centers of excellence
Specialized research facilities:
- Agricultural research stations
- Technical research institutes
- Nuclear research facilities
- Environmental research centers

Sample Terms by Major Languages

English (en)

research institute
think tank
laboratory
documentation center
research center
field station
biological station
research facility
institute of technology
academic department

Spanish (es)

instituto de investigación
centro de investigación
laboratorio
centro de documentación
estación biológica
centro de estudios

German (de)

Forschungsinstitut
Forschungseinrichtung
Biologische Station
Dokumentationszentrum
Forschungszentrum
Forschungslabor

French (fr)

institut de recherche
centre de recherche
laboratoire de recherche
centre de documentation
station biologique

Japanese (ja)

研究所 (kenkyūjo - research institute)
シンクタンク (think tank)
研究センター (research center)
生物実験所 (biological laboratory)

Semantic Breadth Analysis

Hierarchical Structure

The R-class demonstrates significant hierarchical depth:

Top-level: Generic research institutes
Mid-level: Specialized research facility types (labs, stations, centers)
Leaf-level: Institution-specific subtypes (university labs, government research facilities)

Cross-Domain Coverage

Research center types span multiple domains:

Natural Sciences: Biological stations, environmental research centers
Social Sciences: Think tanks, policy research institutes
Technology: Technical research institutes, R&D facilities
Medicine: Medical research institutes, clinical research centers
Agriculture: Agricultural experiment stations
Information Science: Documentation centers, digital humanities centers

Institutional Contexts

University-affiliated research units (institutes, centers, labs)
Government research agencies
Independent research organizations
Corporate R&D facilities
International research collaborations

Data Quality Observations

Strengths

Comprehensive coverage of research center typology
Strong multilingual representation (119 languages)
Rich alternative labels (3,724 total terms for 592 classes)
Good representation of specialized research facility types

Challenges

Overlapping definitions between "research institute", "research center", "laboratory"
Ambiguous boundaries between academic departments and research centers
Some institution-specific subclasses (e.g., specific university institutes) that may be too granular
Mixed levels of abstraction (e.g., "research institute" vs. "Max Planck Institute")

Disambiguation Needs

Laboratory vs. research institute vs. research center
Academic department vs. research unit
Documentation center vs. archive vs. library (overlap with other GLAMORCUBEPSXH classes)
Think tank vs. research institute vs. policy center

Comparison with Other Classes

Relative Metrics

R-class: 592 classes, 3,724 terms, 119 languages
L-class (Library): 364 classes, 2,513 terms, 108 languages
A-class (Archive): 223 classes, 1,794 terms, 91 languages
M-class (Museum): 789 classes, 6,172 terms, 135 languages

Observations:

R-class is moderately sized (larger than A/L, smaller than M)
High term density (6.3 terms/class) indicates rich terminology
Excellent language coverage (119 languages, second only to M-class at 135)

Extraction Methodology

SPARQL Query Design

SELECT DISTINCT ?class ?classLabel ?altLabel WHERE {
  VALUES ?base_class { wd:Q31855 wd:Q7310853 }
  ?class wdt:P279* ?base_class .
  OPTIONAL { ?class skos:altLabel ?altLabel . }
  SERVICE wikibase:label { 
    bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en,nl,de,fr,es,pt,it,ja,zh,ru,ar,hi,ko,vi,th,id,pl,tr,sv,da,no,fi,cs,hu,ro,uk,bg,el,he".
  }
}

Deduplication Process

Input: 3,317 raw SPARQL results
Output: 592 unique classes (82% deduplication rate)
Method: Deduplicated by Wikidata QID, consolidated labels and altLabels per QID

Files Generated

sparql/hyponyms_raw.json - Raw SPARQL results (3,317 records)
sparql/hyponyms_clean.json - Deduplicated results (592 unique classes)
analysis/stats.json - Statistical summary

Next Steps

Immediate Actions

✅ Extract subclass hierarchies from Wikidata
✅ Deduplicate results
✅ Generate initial analysis
⏳ Create controlled vocabulary (top-level terms)
⏳ Build multilingual term mappings
⏳ Document disambiguation rules

Future Enhancements

Map R-class terms to existing research organization taxonomies (e.g., GRID, ROR)
Cross-reference with academic institutional hierarchies
Develop heuristics for distinguishing research centers from academic departments
Create geographic distribution analysis of research center types

Validation Notes

Quality Assurance

Verified base classes exist in Wikidata (Q31855, Q7310853)
Confirmed transitive closure (P279*) captures full hierarchy
Checked sample terms for semantic coherence
Validated language codes against ISO 639-1

Known Issues

Some institution-specific classes may be too granular for vocabulary purposes
Overlap with other GLAMORCUBEPSXH classes (especially L for documentation centers)
Ambiguous boundaries between research center subtypes

Conclusion

The R-class extraction successfully captured a rich, multilingual vocabulary for research centers with 592 unique classes and 3,724 terms across 119 languages. The taxonomy demonstrates broad coverage of research facility types spanning natural sciences, social sciences, technology, and medicine. Key challenges include overlapping definitions and mixed abstraction levels. The dataset provides a solid foundation for building a controlled RESEARCH_CENTER vocabulary.

Extraction Date: 2025-11-12
SPARQL Endpoint: https://query.wikidata.org/sparql
Status: Initial extraction complete ✓

7.1 KiB Raw Blame History