Multilingual Hierarchical Points of Interest (POI) Dataset
Dataset Summary
This dataset contains fully localized Points of Interest (POIs) with deep hierarchical mapping across localized geographic areas and logical operational categories. It natively supports three primary languages representing the Balkan/Eastern Mediterranean region: Greek (EL), English (EN), and Bulgarian (BG).
What makes this dataset distinct is that every POI is enriched not just with its immediate category and sub-region, but also mapped directly to its top-level geographic root node via pre-calculated tree-path evaluations. This enables robust spatial group filtering, semantic cross-lingual search training, and hierarchical multi-label classification modeling.
Supported Tasks
- Tabular & Hierarchical Classification: Training models to predict POI categories or high-level regional distributions based on textual metadata.
- Cross-Lingual Information Retrieval (CLIR): Developing semantic search engines that can query regional names, POI descriptions, and categories across variations of English, Greek, and Cyrillic scripts.
- Geospatial Knowledge Graph Building: Constructing entity relationships mapping
POI -> Category -> Local Area -> Root Area.
Languages
The text data fields are completely localized in:
- Greek (
lang_id=1,gr_prefix) - English (
lang_id=2,en_prefix) - Bulgarian (
lang_id=4,bg_prefix)
Dataset Structure
Data Instance Example
{
"id": 1024,
"gr_name": "Ακρόπολη Αθηνών",
"en_name": "Acropolis of Athens",
"bg_name": "Акропол",
"category_id": 45,
"gr_categoryname": "Αρχαιολογικοί Χώροι",
"en_categoryname": "Archaeological Sites",
"bg_categoryname": "Археологически обекти",
"area_id": 12,
"gr_areaname": "Πλάκα",
"en_areaname": "Plaka",
"bg_areaname": "Плака",
"calculated_root_area_id": 1,
"gr_root_areaname": "Αττική",
"en_root_areaname": "Attica",
"bg_root_areaname": "Αтика"
}
- Downloads last month
- 18