README.md

metadata

language:
  - be
license: cc-by-sa-4.0
task_categories:
  - token-classification
  - text-generation
  - fill-mask
tags:
  - belarusian
  - morphology
  - grammar
  - dictionary
  - lemmatization
  - pos-tagging
pretty_name: GrammarDB (Belarusian Grammatical Database)
size_categories:
  - 1M<n<10M

GrammarDB: Comprehensive Belarusian Grammatical Database

This dataset contains a massive grammatical database of the Belarusian language, featuring approximately 4.5 million entries. It provides exhaustive information on wordforms, their paradigms, morphological features, and accents.

Description

GrammarDB is the primary open-source resource for the morphological analysis of the Belarusian language. It serves as the backbone for systems like LanguageTool, spellcheckers for LibreOffice and Firefox, and various NLP tools. The dataset allows users to map any wordform to its lemma (base form) and a comprehensive set of grammatical categories.

Dataset Structure

Each record in the dataset represents a specific wordform and contains the following fields:

Field	Type	Description
pdgId	int	Paradigm ID (groups wordforms belonging to the same inflectional type).
variantId	char	Paradigm variant identifier.
variantLemma	string	The lemma (base/canonical form) for this variant.
variantType	string?	Variant type (e.g., stylistic or regional variations).
paradigmOptions	string?	Additional paradigm parameters.
regulationType	string?	Normative status (references to specific linguistic rules or dictionaries).
meaning	string?	Brief semantic clarification (used to distinguish homonyms).
theme	string?	Thematic category of the word.
tag	string	Grammatical tags (Part of Speech, gender, case, number, tense, etc.).
slounik	array?	List of source dictionaries where the word is recorded.
pravapis	array?	References to specific orthography (spelling) rules.
formType	string?	Type of the specific form within the paradigm.
formOptions	string?	Form-specific parameters (e.g., stress/accent marks).
form	string	The actual wordform as it appears in text.

For a detailed explanation of the field values and the tagging system, please refer to the official documentation: bnkorpus.info/articles/grammardb.html.

Usage

You can load this dataset using the datasets library:

from datasets import load_dataset

dataset = load_dataset("alex73/GrammarDB")

# Example: Filter all forms for a specific lemma
word_forms = dataset['train'].filter(lambda x: x['variantLemma'] == 'беларускі')
for row in word_forms:
    print(f"{row['form']} -> {row['tag']}")