Hugging Face's logo

Datasets:
vislupus
/
alpaca-bulgarian-dictionary

Tasks:
Text Generation
Modalities:
Text
Formats:
json
Languages:
Bulgarian
Size:
100K - 1M
Libraries:
Datasets
pandas
License:
Dataset card Viewer Files Community
1
alpaca-bulgarian-dictionary / README.md
vislupus's picture
vislupus
Update README.md
1ab5659 verified
preview code
|
raw
history blame contribute delete
3.24 kB
metadata
license: mit
task_categories:
  - text-generation
language:
  - bg
size_categories:
  - 100K<n<1M

Bulgarian Dictionary Dataset

This dataset is a collection of Bulgarian words along with their linguistic details, derived forms, synonyms, incorrect usages, and more. It is designed for use in natural language processing (NLP) tasks, such as training language models, building dictionaries, or enhancing word prediction systems.

Source Information

The data is sourced from the Читанка Речник, a free online dictionary. The Читанка Речник project aims to preserve and make accessible the richness of the Bulgarian language.

Dataset Description

The dataset contains detailed linguistic information about Bulgarian words, formatted for NLP tasks. Each word entry includes the following fields (when available):

Fields:

  1. Word (name): The Bulgarian word.
  2. Stressed Form (name_stressed): The word with its stress pattern marked.
  3. Meaning (meaning): A description of the word's definition or its usage.
  4. Synonyms (synonyms): Other words with similar meanings.
  5. Incorrect Forms (incorrect_forms): Common misspellings or misused forms of the word.
  6. Derivative Forms (derivative_forms): Related forms of the word, such as conjugations or declensions.
  7. Broken Form (name_broken): A segmented version of the word to indicate its morphological structure.

Format

The dataset is formatted in JSON for compatibility with a wide range of applications, including:

  • Training large language models (e.g., GPT, BERT).
  • Fine-tuning for dictionary-like querying.
  • Building language tools like spelling correctors or synonym finders.

Key Features

  1. Rich Linguistic Annotations:

    • Includes stress patterns, morphological segmentation, and derivative forms.
    • Captures common errors for spelling correction.

  2. Diverse Prompts for NLP Training:

    • Includes formal and informal instructions for tasks like querying meanings, synonyms, and more.
    • Offers 20+ unique phrasing styles for each query type, enhancing model robustness.

  3. Comprehensive Coverage:

    • Designed to reflect the richness and complexity of the Bulgarian language.

Usage

The dataset can be used in various NLP applications, such as:

  • Training conversational AI models to respond to dictionary-like queries.
  • Building tools for Bulgarian spelling correction and synonym generation.
  • Enhancing text prediction and completion for Bulgarian text.

Acknowledgments

This dataset is based on information provided by Читанка Речник. The original project plays a vital role in preserving and promoting the Bulgarian language.

Citation

If you use this dataset in your work, please credit Читанка Речник and provide a link to their website: Читанка Речник