README.md

---
license: gfdl
task_categories:
- text-classification
- token-classification
language:
- es
tags:
- spanish
- morphology
- lemmatization
- wiktionary
- nlp
- linguistics
size_categories:
- 100K<n<1M
dataset_info:
  features:
  - name: word
    dtype: string
  - name: original form
    dtype: string
  config_name: default
  splits:
  - name: train
    num_bytes: 24985600
    num_examples: 760159
  - name: only_adj_noun_verb
  download_size: 12492800
  dataset_size: 24985600
configs:
- config_name: default
  data_files:
  - split: train
    path: train.csv
  - split: only_adj_noun_verb
    path: es_word_forms_only_adj.noun.verb.csv.csv
---

# Spanish Word Forms and Lemmas Dataset

## Dataset Description

This dataset contains Spanish word forms paired with their corresponding lemmas (original forms), extracted from Wiktionary. It provides a comprehensive mapping of inflected Spanish words to their base forms, making it valuable for morphological analysis, lemmatization tasks, and Spanish NLP applications.

### Dataset Summary

- **Language**: Spanish (es)
- **Task**: Lemmatization, Morphological Analysis
- **Source**: Wiktionary
- **Size**: ~760K word form pairs
- **Format**: CSV with two columns: `word` (inflected form) and `original_form` (lemma)

### Supported Tasks

- **Lemmatization**: Map inflected words to their base forms
- **Morphological Analysis**: Study Spanish word inflection patterns
- **Text Preprocessing**: Normalize Spanish text for NLP pipelines
- **Language Model Training**: Pre-training data for Spanish morphology understanding

## Dataset Structure

### Data Instances

Each instance contains:
- `word`: An inflected Spanish word form
- `original_form`: The corresponding lemma or base form

Example:
```json
{
  "word": "adoquinás",
  "original_form": "adoquinar"
}
```

### Data Fields

- `word` (string): The inflected or conjugated form of a Spanish word
- `original_form` (string): The lemma or dictionary form of the word

### Data Splits

| Split | Examples |
|-------|----------|
| train | 760,159  |

## Dataset Creation

### Source Data

This dataset was extracted from Wiktionary, specifically focusing on Spanish language entries. Wiktionary is a collaborative, multilingual dictionary project that provides comprehensive morphological information for words across many languages.

### Data Collection Process

The data was systematically extracted from Wiktionary's Spanish language entries, capturing the relationships between inflected word forms and their corresponding lemmas. This includes:

- Verb conjugations (all tenses, moods, and persons)
- Noun and adjective inflections (gender and number variations)
- Diminutives, augmentatives, and other morphological variations

### Data Quality

- The dataset provides high-quality mappings based on Wiktionary's editorial standards
- Each word form is mapped to its appropriate lemma
- Covers a wide range of Spanish morphological patterns

## Uses

### Direct Use

This dataset can be directly used for:
- Training lemmatization models for Spanish
- Building morphological analyzers
- Creating Spanish text preprocessing pipelines
- Studying Spanish morphological patterns

### Downstream Tasks

- **Named Entity Recognition**: Improve NER by normalizing word forms
- **Information Retrieval**: Enhance search by matching different word forms
- **Machine Translation**: Better handling of Spanish morphology
- **Text Classification**: Normalize features for better classification performance

## Considerations for Use

### Social Impact of Dataset

This dataset supports the development of Spanish NLP tools, which can:
- Improve accessibility of Spanish-language technology
- Support educational applications for Spanish learners
- Enable better processing of Spanish text in various domains

### Limitations

- Limited to words present in Wiktionary (may not include very recent terms or highly specialized vocabulary)
- Focuses on standard Spanish forms (may not capture all dialectal variations)
- Does not include semantic information beyond morphological relationships

## Additional Information

### Dataset Curators

This dataset was curated from Wiktionary data focusing on Spanish morphological relationships.

### Licensing Information

This dataset is released under the MIT License, following Wiktionary's open data principles.

### Citation Information

If you use this dataset in your research, please cite:

```bibtex
@dataset{spanish_word_forms_wiktionary,
  title={Spanish Word Forms and Lemmas Dataset from Wiktionary},
  author={Wiktionary Contributors},
  year={2024},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/your-username/spanish-word-forms-wiktionary}
}
```

### Contributions

This dataset was created by extracting and processing data from Wiktionary. Thanks to all Wiktionary contributors who have provided the underlying linguistic data.