The CongoNames Corpus: A Large-Scale Labeled Dataset of Congolese Personal Names

---

Abstract

Personal names are fundamental to cultural identity, yet many African countries lack large, structured datasets of names for language research and technology. This paper presents CONGONAMES, the first large-scale corpus of personal names from the Democratic Republic of the Congo (DRC). Derived from public national exam records, the corpus contains over 8 million name entries standardized and enriched with metadata including gender and regional provenance. We detail a reproducible pipeline for extracting and normalizing name data from PDF documents using automated parsing and regex-based formatting. The resulting dataset provides rich opportunities to analyze naming conventions across one of Africa’s most linguistically diverse nations. We benchmark basic statistics of the corpus and illustrate its potential use in downstream applications. By releasing the corpus and processing tools openly, we aim to support research in African NLP, onomastics, and social science, and to address the gap in resources for Congolese and African personal names.

Workflows

Clone the repository and sync dependencies:

git clone https://github.com/bernard-ng/drc-names-corpus.git
cd drc-names-corpus
uv sync 

Run the following scripts in order:

uv run drc-names-corpus collect     # Download PDFs from the data source.
uv run drc-names-corpus extract     # Extract text files from downloaded PDFs.   
uv run drc-names-corpus format      # Normalize and format extracted text files.

Final workflow sequence: collect -> extract -> format -> [export names, ablation, export features, export components, export unique, export statistics] -> report

Exporting

Use the unified export dispatcher to generate dataset artifacts in the final workflow order:

uv run drc-names-corpus export names
uv run drc-names-corpus ablation
uv run drc-names-corpus export features
uv run drc-names-corpus export components
uv run drc-names-corpus export unique
uv run drc-names-corpus export statistics
uv run drc-names-corpus report

Dataset Exports

All exported CSVs are written under dataset/gold/ and normalized to lowercase strings.

dataset/gold/names.csv

Column	Type	Description
id	string	Candidate identifier, left-padded to three digits.
name	string	Full candidate name as recorded in the source text.
sex	string	Reported sex marker (m or f).
year	string	Exam year parsed from the filename.
region	string	Region label parsed from the filename.
filename	string	Source text filename.
line	integer	Line number in the gold text file where the entry appears.

dataset/gold/statistics.csv

Column	Type	Description
index	integer	Row index within the source file.
name	string	School name as recorded in the source text.
code	string	School code (normalized to the expected numeric format).
entries	integer	Total candidates listed for the school.
pass	integer	Candidates who passed.
fail	integer	Candidates who failed.
entries_f	integer	Female candidates listed.
entries_m	integer	Male candidates listed.
pass_f	integer	Female candidates who passed.
pass_m	integer	Male candidates who passed.
fail_f	integer	Female candidates who failed.
fail_m	integer	Male candidates who failed.
year	string	Exam year parsed from the filename.
region	string	Region label parsed from the filename.
filename	string	Source text filename.

dataset/gold/names_featured.csv

Column	Type	Description
id	string	Candidate identifier, left-padded to three digits.
name	string	Full candidate name as recorded in the source text.
sex	string	Reported sex marker (m or f).
year	string	Exam year parsed from the filename.
region	string	Region label parsed from the filename.
filename	string	Source text filename.
line	integer	Line number in the gold text file where the entry appears.
words	integer	Word count in the name.
length	integer	Character count in the name (including spaces).
category	string	simple if the name has exactly three words, otherwise complex.
province	string	Province inferred from region via the lookup map.

dataset/gold/names_unique.csv

Column	Type	Description
id	string	Candidate identifier for the first occurrence of the name.
name	string	Unique full name (deduplicated).
sex	string	Reported sex marker (m or f).
year	string	Exam year parsed from the filename.
region	string	Region label parsed from the filename.
filename	string	Source text filename.
line	integer	Line number for the first occurrence.

dataset/gold/names_components.csv

Column	Type	Description
id	string	Candidate identifier for the original full name.
name	string	Full name associated with the component.
sex	string	Reported sex marker (m or f).
year	string	Exam year parsed from the filename.
region	string	Region label parsed from the filename.
filename	string	Source text filename.
line	integer	Line number in the gold text file where the entry appears.
full_name	string	Original full name for the component.
component	string	One component extracted from the full name.

dataset/gold/names_unstructured.csv

Column	Type	Description
name	string	Extracted unstructured name string from ablation text.
filename	string	Source ablation text filename.
line	integer	Line number in the ablation text file.