Methodology

Data Engineering Framework for Medieval Institutional Analysis

Author

Nambona YANGUERE

Published

April 3, 2026

1. Research Objective

This study aims to quantify the institutional, economic, and geopolitical impact of medieval monastic communities across Europe using modern data engineering and ESG analysis frameworks.

The research addresses three questions:

How did different religious orders (as institutional protocols) influence economic output and territorial development?
What is the correlation between governance structure and institutional resilience across geopolitical disruptions?
Can we quantify the environmental and cultural legacy of monastic land management using ESG metrics?

2. Data Acquisition

2.1 Sources

Source	Type	Content
Ecclesiastical registers	Archival	Foundation dates, order affiliation, patronage
Medieval cartularies	Archival	Land grants, tithes, economic transactions
Archaeological surveys	Geospatial	Site coordinates, architectural remains
Trade route databases	Geospatial	Medieval commercial corridors
Academic publications	Secondary	Institutional history, regional economics

2.2 Collection Method

Data was collected through a combination of:

Automated web scraping (Python / BeautifulSoup) targeting digitized archival databases and open-access historical repositories
LLM-assisted Named Entity Recognition (NER) to extract structured entities (founders, orders, dates, land grants) from unstructured historical text in Latin and Old French
Manual verification against published academic sources to validate AI-extracted entities

3. Data Normalization

3.1 Standardization

Raw data from heterogeneous sources required normalization across several dimensions:

Dimension	Method	Tool
Dates	Conversion to ISO 8601 from Latin calendar references	Python (dateutil)
Locations	Geocoding historical place names to modern coordinates	GeoPandas / Nominatim
Economic values	Min-Max scaling of tithes and land grants	Pandas / Scikit-learn
Order names	Mapping variant spellings to canonical order names	Custom lookup table
Text language	Translation of Latin/Old French excerpts	LLM-assisted

3.2 Dataset Structure

After normalization, five structured datasets were produced:

4. Analytical Framework

4.1 ESG Scoring

Each community receives a composite score (0–100) across three dimensions:

Dimension	Weight	Indicators
Environmental	35%	Land reclamation, agricultural diversification, water management
Social	35%	Education, hospitality, community employment, cultural production
Governance	30%	Rule adherence, institutional continuity, succession protocols

4.2 Geopolitical Analysis

Communities are analyzed against:

Proximity to trade routes (km to nearest documented corridor)
Border zone classification (frontier vs. interior positioning)
Political stability index of the surrounding region per century
Order-level territorial strategy (documented expansion protocols)

4.3 Network Analysis

Sister communities are linked using:

Same order affiliation within a defined radius (150 km)
Documented dependency relationships (mother/daughter houses)
Shared patronage networks (same founding family or bishop)

4.4 Temporal Analysis

Economic and institutional data is segmented into century-level periods to identify:

Growth phases correlated with political stability
Decline phases correlated with geopolitical disruption
Resilience patterns linked to governance protocols

5. Reproducibility

All code is available in the project repository:

src/scraping.py — Data acquisition pipeline
src/ner_extraction.py — AI entity extraction
src/normalization.py — Cleaning and scaling
src/esg_scoring.py — ESG dimension calculations
src/network.py — Sister community proximity
src/viz.py — Visualization functions

Raw data is excluded from the repository (.gitignore) to respect archival licensing. Processed datasets are included in data/processed/ for full reproducibility of the analysis.

References

--- title: "Methodology" subtitle: "Data Engineering Framework for Medieval Institutional Analysis" author: "Nambona YANGUERE" date: last-modified format: html: css: report-style.css toc: true toc-depth: 3 code-fold: true bibliography: references.bib --- ## 1. Research Objective This study aims to quantify the institutional, economic, and geopolitical impact of medieval monastic communities across Europe using modern data engineering and ESG analysis frameworks. The research addresses three questions: 1. How did different **religious orders** (as institutional protocols) influence economic output and territorial development? 2. What is the correlation between **governance structure** and institutional **resilience** across geopolitical disruptions? 3. Can we quantify the **environmental and cultural legacy** of monastic land management using ESG metrics? ## 2. Data Acquisition ### 2.1 Sources | Source | Type | Content | | :--- | :--- | :--- | | Ecclesiastical registers | Archival | Foundation dates, order affiliation, patronage | | Medieval cartularies | Archival | Land grants, tithes, economic transactions | | Archaeological surveys | Geospatial | Site coordinates, architectural remains | | Trade route databases | Geospatial | Medieval commercial corridors | | Academic publications | Secondary | Institutional history, regional economics | ### 2.2 Collection Method Data was collected through a combination of: - **Automated web scraping** (Python / BeautifulSoup) targeting digitized archival databases and open-access historical repositories - **LLM-assisted Named Entity Recognition (NER)** to extract structured entities (founders, orders, dates, land grants) from unstructured historical text in Latin and Old French - **Manual verification** against published academic sources to validate AI-extracted entities ```{=python} #| echo: true #| label: code-ner #| code-summary: "NER Extraction Example" # Simplified example of entity extraction pipeline # Full implementation in src/ner_extraction.py import json def extract_entities(text: str) -> dict: """ Extract structured entities from historical text. Uses LLM-based NER for Latin/Old French sources. Returns: dict with keys: name, order, founded, location, patron """ # LLM prompt template for entity extraction prompt = f""" Extract the following entities from this historical text: - Community name - Religious order - Foundation date - Geographic location - Founding patron (if mentioned) Text: {text} Return as JSON. """ # result = llm.generate(prompt) # return json.loads(result) pass ``` ## 3. Data Normalization ### 3.1 Standardization Raw data from heterogeneous sources required normalization across several dimensions: | Dimension | Method | Tool | | :--- | :--- | :--- | | Dates | Conversion to ISO 8601 from Latin calendar references | Python (dateutil) | | Locations | Geocoding historical place names to modern coordinates | GeoPandas / Nominatim | | Economic values | Min-Max scaling of tithes and land grants | Pandas / Scikit-learn | | Order names | Mapping variant spellings to canonical order names | Custom lookup table | | Text language | Translation of Latin/Old French excerpts | LLM-assisted | ### 3.2 Dataset Structure After normalization, five structured datasets were produced: ```{=python} #| echo: true #| label: code-schema #| code-summary: "Dataset Schemas" schemas = { "monastic_sites.csv": [ "name", "order", "founded", "location", "lat", "lon", "status", "dissolution_cause", "peak_year", "decline_year", "dissolution_year", "architectural_legacy", "cultural_legacy" ], "orders_metadata.csv": [ "order", "governance", "economic_model", "territorial_strategy", "order_founded" ], "esg_metrics.csv": [ "name", "environmental", "social", "governance", "composite" ], "community_network.csv": [ "source", "target", "distance_km", "shared_period" ], "economic_indicators.csv": [ "name", "year", "revenue_index", "land_area", "primary_revenue_source" ] } for dataset, columns in schemas.items(): print(f"{dataset}: {len(columns)} columns") ``` ## 4. Analytical Framework ### 4.1 ESG Scoring Each community receives a composite score (0–100) across three dimensions: | Dimension | Weight | Indicators | | :--- | :--- | :--- | | Environmental | 35% | Land reclamation, agricultural diversification, water management | | Social | 35% | Education, hospitality, community employment, cultural production | | Governance | 30% | Rule adherence, institutional continuity, succession protocols | ### 4.2 Geopolitical Analysis Communities are analyzed against: - **Proximity to trade routes** (km to nearest documented corridor) - **Border zone classification** (frontier vs. interior positioning) - **Political stability index** of the surrounding region per century - **Order-level territorial strategy** (documented expansion protocols) ### 4.3 Network Analysis Sister communities are linked using: - **Same order affiliation** within a defined radius (150 km) - **Documented dependency relationships** (mother/daughter houses) - **Shared patronage networks** (same founding family or bishop) ### 4.4 Temporal Analysis Economic and institutional data is segmented into century-level periods to identify: - **Growth phases** correlated with political stability - **Decline phases** correlated with geopolitical disruption - **Resilience patterns** linked to governance protocols ## 5. Reproducibility All code is available in the project repository: - `src/scraping.py` — Data acquisition pipeline - `src/ner_extraction.py` — AI entity extraction - `src/normalization.py` — Cleaning and scaling - `src/esg_scoring.py` — ESG dimension calculations - `src/network.py` — Sister community proximity - `src/viz.py` — Visualization functions Raw data is excluded from the repository (`.gitignore`) to respect archival licensing. Processed datasets are included in `data/processed/` for full reproducibility of the analysis. ## References