metadata

license: cc-by-nc-sa-4.0
language:
  - en
  - my
pretty_name: Myanmar Bilingual Village Address Directory (72k+ Entries)
tags:
  - myanmar
  - bilingual
  - geospatial
  - address-parsing
  - named-entity-recognition
  - translation

Myanmar Bilingual Village Address Directory (72k+ Entries)

Dataset Description

This dataset provides a comprehensive, bilingual (English and Myanmar) list of over 72,000 rural village addresses across Myanmar. Each entry is a clean, human-readable address string containing the village, village tract, township, district, and state/region.

The dataset is ideal for a wide range of NLP and data science tasks, including Named Entity Recognition (NER), address parsing, machine translation, and as a foundational resource for logistics and geospatial analysis in the Myanmar context.

This final, cleaned dataset was prepared by freococo.

Data Fields

The dataset contains a single CSV file (myanmar_village_based_72k_addresses.csv) with two columns, representing a parallel corpus of addresses.

Address_English: The complete address string in English.
Address_Myanmar: The complete address string in Myanmar script.

Example Row:

Address_English	Address_Myanmar
`Da None Chaung, (Kyun Nyo Gyi) Kyun Hteik, Bogale Township, Pyapon District, Ayeyarwady`	`ဓနုံးချောင်း, (ကျွန်းညိုကြီး) ကျွန်းထိပ် ကျေးရွာအုပ်စု, ဘိုကလေး မြို့နယ်, ဖျာပုံခရိုင်, ဧရာဝတီတိုင်းဒေသကြီး`

Data Splits

The dataset consists of a single split, train, containing 72,569 entries.

Dataset Creation

The dataset was derived from the Myanmar Place Codes (P-Codes) Release 9.6 (Feb 2025), provided by the Myanmar Information Management Unit (MIMU), a service of the United Nations in Myanmar. The source data was downloaded from the Humanitarian Data Exchange (HDX).

The original hierarchical files were programmatically merged, de-duplicated, and transformed into this clean, flat address list using Python and the pandas library.

Usage

The dataset can be easily loaded using the Hugging Face datasets library.

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("freococo/myanmar_village_based_72k_addresses")

# Access an example from the dataset
print(dataset['train'])

# Expected output:
# {
#  'Address_English': 'Da None Chaung, (Kyun Nyo Gyi) Kyun Hteik, Bogale Township, Pyapon District, Ayeyarwady',
#  'Address_Myanmar': 'ဓနုံးချောင်း, (ကျွန်းညိုကြီး) ကျွန်းထိပ် ကျေးရွာအုပ်စု, ဘိုကလေး မြို့နယ်, ဖျာပုံခရိုင်, ဧရာဝတီတိုင်းဒေသကြီး'
# }

Use Cases and Limitations

Potential Use Cases

Named Entity Recognition (NER): This dataset is a valuable resource for training models to recognize Myanmar place names (e.g., B-STATE, I-DISTRICT, I-VILLAGE).
Address Parsing: It can be used to train models that can take an unstructured address string and break it down into its constituent administrative parts (village, township, etc.).
Machine Translation: The bilingual format makes it a useful parallel corpus for fine-tuning translation models specifically for Myanmar administrative terms.
Data Augmentation: Can be used to create realistic synthetic data for applications requiring Myanmar addresses.

Limitations

Rural Only: This dataset exclusively contains rural village addresses and does not include urban administrative units like towns or wards.
No Geographic Coordinates: The final formatted file does not contain latitude and longitude data. For geocoding, this dataset would need to be joined back to the original source files.
Static Data: The dataset is based on the February 2025 release from MIMU and will not reflect any subsequent administrative changes.

Citation Information

If you use this dataset in your work, please cite the original data source and acknowledge the creator of this specific version.

Citing the Original Source

Myanmar Information Management Unit (MIMU). (2025). MIMU Place Codes (P-Codes) Release 9.6. Sourced from the Humanitarian Data Exchange (HDX).

Citing this Dataset Repository

@dataset{freococo_myanmar_village_addresses,
  author = {freococo},
  title = {Myanmar Bilingual Village Address Directory (72k+ Entries)},
  year = {2025},
  publisher = {Hugging Face},
  version = {1.0.0},
  url = {https://huggingface.co/datasets/freococo/myanmar_village_based_72k_addresses}
}