bettergovph/raw-philippine-data

Raw Philippine Data

This repository contains raw data about Philippine politicians, public officials, and legislative documents collected from various sources. The data is intended for research, analysis, and civic technology purposes.

Dataset Overview

This dataset currently contains:

Persons

45,424 person records of Philippine politicians and public officials with:

• ID: Unique identifier (ULID format)
• First Name: Person's first name
• Last Name: Person's last name
• Name Suffix: Jr., Sr., I, II, III, IV, etc. (if applicable)

Memberships

Political party affiliations and positions held by persons, including:

• ID: Unique membership identifier
• Person ID: Links to the person record
• Party: Political party affiliation
• Region: Geographic region (e.g., "National Capital Region", "Region III")
• Province: Province name
• Locality: City or municipality (optional)
• Position: Position held (e.g., "Representative", "Governor", "Mayor")
• Year: Year of the position/membership

Documents

60,934 legislative documents including Senate Bills (SB) and House Bills (HB) from various Congressional sessions:

• ID: Unique document identifier (e.g., "sb-20-2" for Senate Bill 2 from 20th Congress)
• Document Type: Type of document ("sb" for Senate Bill, "hb" for House Bill)
• Congress: Congressional session number (e.g., 17, 18, 19, 20)
• Document Number: Official bill/document number
• File Path: Path to the source text file
• Content: Full text content of the document

More entity types (groups, etc.) will be added in the future.

Using the Dataset

Browse in Hugging Face Dataset Viewer

You can explore the data directly in your browser using the Dataset Viewer tab above.

• Select "persons" from the config dropdown to view person records
• Select "memberships" to view political positions and party affiliations
• Select "documents" to view legislative bills and documents
• Additional entity types will appear in the dropdown as they're added

The data is available in Parquet format for easy viewing and filtering.

Load with Hugging Face Datasets

from datasets import load_dataset

# Load persons data
persons = load_dataset("bettergovph/raw-philippine-data", "persons")
print(persons['train'][0])

# Load memberships data
memberships = load_dataset("bettergovph/raw-philippine-data", "memberships")
print(memberships['train'][0])

# Load documents data
documents = load_dataset("bettergovph/raw-philippine-data", "documents")
print(documents['train'][0])

# Future: Load other entity types
# groups = load_dataset("bettergovph/raw-philippine-data", "groups")

Query with DuckDB

For advanced SQL queries, download the DuckDB database:

git clone https://huggingface.co/datasets/bettergovph/raw-philippine-data
cd raw-philippine-data
duckdb databases/data.duckdb

Example queries:

-- Count all persons
SELECT COUNT(*) FROM persons;

-- Count all memberships
SELECT COUNT(*) FROM memberships;

-- Count all documents
SELECT COUNT(*) FROM documents;

-- Find all persons with "Jr." suffix
SELECT * FROM persons WHERE name_suffix = 'Jr.' LIMIT 10;

-- Search by last name
SELECT * FROM persons WHERE last_name LIKE 'Aquino%';

-- Group by name suffix
SELECT name_suffix, COUNT(*) as count
FROM persons
WHERE name_suffix IS NOT NULL
GROUP BY name_suffix
ORDER BY count DESC;

-- Find all mayors in a specific region
SELECT p.first_name, p.last_name, m.province, m.locality, m.year
FROM memberships m
JOIN persons p ON m.person_id = p.id
WHERE m.position = 'Mayor'
  AND m.region = 'National Capital Region'
ORDER BY m.year DESC
LIMIT 10;

-- Count positions by party affiliation
SELECT party, position, COUNT(*) as count
FROM memberships
WHERE party IS NOT NULL
GROUP BY party, position
ORDER BY count DESC
LIMIT 20;

-- Find persons with multiple political positions
SELECT p.first_name, p.last_name, COUNT(*) as position_count
FROM persons p
JOIN memberships m ON p.id = m.person_id
GROUP BY p.id, p.first_name, p.last_name
HAVING COUNT(*) > 1
ORDER BY position_count DESC
LIMIT 10;

-- Search documents by keyword in content
SELECT id, document_type, congress, document_number, LENGTH(content) as content_length
FROM documents
WHERE content LIKE '%infrastructure%'
LIMIT 10;

-- Count documents by type and congress
SELECT document_type, congress, COUNT(*) as count
FROM documents
GROUP BY document_type, congress
ORDER BY congress DESC, document_type;

-- Find a specific Senate Bill
SELECT id, congress, document_number, SUBSTR(content, 1, 200) as preview
FROM documents
WHERE document_type = 'sb' AND congress = 20 AND document_number = 2;

Data Sources

The raw data comes from multiple sources:

• Persons & Memberships: TOML files in the data/person/ directory. Each person has their own TOML file with their information, including an optional memberships array that contains their political positions and party affiliations.

• Documents: Text files in the data/document/ directory, organized by document type (sb/hb), congress number, and document ranges. For example:

  • data/document/sb/20/00001-01000/SB-00002.txt - Senate Bill 2 from the 20th Congress
  • data/document/hb/20/04001-05000/HB-04321.txt - House Bill 4321 from the 20th Congress

Regenerating the Dataset

If you've made changes to the source data files and want to regenerate the database and Parquet files:

# Install dependencies
pip install -r requirements.txt

# Load persons data and export to Parquet
python scripts/load_persons_to_db.py --export-parquet

# Load documents data and export to Parquet
python scripts/load_documents_to_db.py --export-parquet

# Optional: Use larger batch size for faster loading
python scripts/load_persons_to_db.py --export-parquet --batch-size 5000
python scripts/load_documents_to_db.py --export-parquet --batch-size 5000

This will create:

• databases/data.duckdb - DuckDB database for SQL queries
• databases/persons.parquet - Persons table in Parquet format
• databases/memberships.parquet - Memberships table in Parquet format
• databases/documents.parquet - Documents table in Parquet format

The scripts use batch inserts for performance and include:

• Progress tracking with percentage complete
• Error logging to databases/load_*_errors.log
• Total execution time reporting
• Graceful handling of Ctrl+C interruptions
• Sample data preview and statistics

Note: Future entity types (groups, etc.) will also generate their own parquet files in the databases/ folder.

Contributing

Contributions are welcome! You can help by:

• Adding new person records (create TOML files in data/person/)
• Adding new legislative documents (add text files in data/document/)
• Updating existing records with more information
• Reporting data quality issues
• Improving documentation

Impostor Syndrome Disclaimer

We want your help. No, really.

There may be a little voice inside your head that is telling you that you're not ready to be an open source contributor; that your skills aren't nearly good enough to contribute. What could you possibly offer a project like this one?

We assure you - the little voice in your head is wrong. If you can write code at all, you can contribute code to open source. Contributing to open source projects is a fantastic way to advance one's coding skills. Writing perfect code isn't the measure of a good developer (that would disqualify all of us!); it's trying to create something, making mistakes, and learning from those mistakes. That's how we all improve, and we are happy to help others learn.

Being an open source contributor doesn't just mean writing code, either. You can help out by writing documentation, tests, or even giving feedback about the project (and yes - that includes giving feedback about the contribution process). Some of these contributions may be the most valuable to the project as a whole, because you're coming to the project with fresh eyes, so you can see the errors and assumptions that seasoned contributors have glossed over.

Remember:

• No contribution is too small
• Everyone started somewhere
• Questions are welcome
• Mistakes are learning opportunities
• Your perspective is valuable

(Impostor syndrome disclaimer adapted from Adrienne Friend)

License

This dataset is licensed under the CC0 1.0 Universal license. This means you can copy, modify, distribute and perform the work, even for commercial purposes, all without asking permission.