Dataset Card for Dataset oasolisr/crPoliticalCorpus2026

Dataset Summary

is a curated collection of political text paragraphs extracted from multiple PDF sources of Costa Rican political parties.
Each document is segmented into clean, well-structured paragraphs using OCR, regex normalization, and paragraph-level filtering.

The dataset is intended for research on political text classification, RAG pipelines, ideological modeling, document similarity, and NLP experiments in Spanish language corpora.

Supported Tasks and Benchmarks

• Retrieval-Augmented Generation (RAG)
• Summarization
• Document Clustering

Languages

• Spanish (es)
  Extracted from Costa Rican political documents.

Dataset Structure

Data Fields

Each row in the dataset corresponds to one paragraph from a political document, with the following fields:

• doc_id (int) – Unique ID for each source document
• title (string) – Name of the original document
• text (string) – Cleaned paragraph text

Example Row

{
  "doc_id": 3,
  "title": "PUSC",
  "text": "Costa Rica faces a decisive moment: its democracy, social cohesion, and institutional capacity are under pressure..."
}