alex73/GrammarDB

---
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](https://bnkorpus.info/articles/grammardb.html).

## Usage

You can load this dataset using the `datasets` library:

```python
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']}")