theses-fr - SKILL.md Agent Skill

name: theses-fr description: 500,000+ French doctoral theses from the national theses.fr database, covering all disciplines since 1985. Titles, authors, directors, institutions, disciplines, defense dates, subject keywords (free-text and RAMEAU controlled vocabulary), jury members. Use for any question about French higher education trends, academic research output, doctoral workforce, or scientific discipline evolution in France.

French National Doctoral Theses Database (theses.fr)

SDK that provides access to the French national database of doctoral theses defended and in preparation since 1985. Covers 500,000+ thesis records with rich metadata including authors, directors, institutions, disciplines, defense dates, and subject keywords. Useful for tracking trends in French academic research, higher education policy, and scientific workforce development.

Quick Start

from sdk_theses_fr import Client

client = Client()

Key Methods

Method	What it does
`client.theses.search(...)`	Search theses with rich filtering (discipline, institution, author, director, subject, date range, language, status)
`client.theses.count(...)`	Count matching theses without fetching full results -- efficient for trend analysis
`client.theses.get_by_nnt(nnt)`	Look up a specific thesis by its National Thesis Number
`client.personnes.search(...)`	Search persons associated with theses (authors, directors, jury members)
`client.personnes.count(...)`	Count matching persons

Data Coverage

Domain: Education / Research / Science
Countries/Regions: France
Time range: 1985 to present
Update frequency: Daily (new defenses registered continuously)
Key indicators: Thesis defenses per year, disciplines, institutions, doctoral schools, research topics, thesis directors, jury composition

Forecasting Patterns

Trend analysis: Use theses.count() with year-by-year date ranges to build time series of thesis defenses by discipline, institution, or status. Track growth/decline in specific research fields.
Point-in-time data: All search methods support cutoff_date parameter to get data as it existed at a specific date. Use for accurate historical comparisons without look-ahead bias.
Entity comparison: Compare thesis output across institutions (institution="Sorbonne" vs institution="Polytechnique"), disciplines, or doctoral schools to identify relative strengths and trends.
Weak signal detection: Search by subject keywords to find emerging research topics before they become mainstream disciplines. Track new doctoral schools or institutions appearing in the database.
Cross-reference: Combine with INSEE economic data for correlating research investment with thesis output, or with labor market data to track doctoral workforce supply.

Common Queries

# Count theses defended in a given year by discipline
count_cs_2023 = client.theses.count(
    discipline="Informatique",
    date_soutenance_start="2023-01-01",
    date_soutenance_end="2023-12-31",
)

# Search theses on a specific topic
ai_theses = client.theses.search(
    subject="intelligence artificielle",
    status="soutenue",
    nombre=50,
)

# Find a specific researcher's thesis supervision history
supervisors = client.personnes.search(nom="Piketty")

# Track in-preparation theses (leading indicator of future defenses)
in_prep = client.theses.search(status="enCours", discipline="Physique", nombre=10)

# Use pagination for large result sets
page1 = client.theses.search(discipline="Droit", nombre=100, debut=0)
page2 = client.theses.search(discipline="Droit", nombre=100, debut=100)

Full Method Reference

See references/methods.md for all 5 methods with complete parameter details.