OurNakshatra/ournakshatra-vedic-astrology-core

OurNakshatra Vedic Astrology Core Dataset

Dataset Summary

The ournakshatra-vedic-astrology-core dataset is a highly structured, expert-curated collection of 602 Q&A pairs covering foundational and advanced concepts in Vedic Astrology (Jyotish). It was developed by the team at OurNakshatra to address the severe lack of high-quality, hallucination-free Vedic astrology training data available to the open-source AI community.

Modern language models frequently struggle with Vedic astrology, often hallucinating calculations, conflating the sidereal (Vedic) and tropical (Western) zodiacs, or misrepresenting classical texts like the Brihat Parashara Hora Shastra (BPHS).

This dataset serves as a canonical Ground Truth repository, providing clean, accurately translated, and culturally contextualized definitions for:

• All 9 Grahas (Planets) and their significations.
• All 27 Nakshatras (Lunar Mansions) and their characteristics.
• All 12 Bhavas (Houses) and their domains.
• Complex astrological mechanics including Combustion, Retrograde motion, Exaltation, and Doshas.

Supported Tasks and Leaderboards

• question-answering: The dataset can be used to train models for open-domain QA on Vedic astrology.
• text-generation: Can be used to fine-tune models to generate accurate, non-hallucinated astrological interpretations.
• RAG Systems: Highly suitable for Retrieval-Augmented Generation pipelines requiring authoritative astrological definitions.

Languages

The text in the dataset is in English (en), featuring standardized phonetic transliterations of Sanskrit terminology where appropriate.

Dataset Structure

Data Instances

A typical instance consists of a question, a comprehensive answer, and the source attribution.

{
  "question": "What is Lagna in Vedic Astrology?",
  "answer": "The zodiac sign rising over the eastern horizon at the exact moment of birth. The foundation of the entire chart \u2014 every house is numbered from the Lagna.",
  "source": "OurNakshatra.com"
}

Data Fields

• question: (string) The astrological concept or mechanical query.
• answer: (string) The detailed, expert-verified explanation of the concept, stripped of HTML formatting.
• source: (string) The attribution source (OurNakshatra.com).

Data Splits

The dataset currently consists of a single train split containing 602 high-quality examples.

Dataset Creation

Curation Rationale

AI researchers often rely on messy web scrapes for esoteric subjects like astrology, resulting in models that produce inaccurate or culturally confused outputs. OurNakshatra curated this dataset to inject structural integrity and classical accuracy into the AI ecosystem, ensuring that models trained on this data will respect the mathematical and philosophical foundations of Jyotish.

Source Data

The data was extracted directly from the proprietary content engine powering OurNakshatra.com, written and verified by practitioners of classical Vedic astrology.

Data Cleaning and Formatting

The raw content was processed to remove all HTML tags, styling, and application-specific metadata. The resulting text is clean, UTF-8 encoded plain text optimized for LLM ingestion and MinHash deduplication pipelines.

Considerations for Using the Data

Social Impact of Dataset

By providing accurate cultural and mathematical definitions of an ancient knowledge system, this dataset helps prevent the algorithmic distortion of traditional Indian astronomy and philosophy by global AI models.

Limitations

This dataset provides conceptual definitions and mechanical rules. It does not provide real-time Swiss Ephemeris data or coordinate math. Models trained on this dataset should be configured to route dynamic calculation requests (e.g., "Calculate my Kundali") to dedicated ephemeris APIs or platforms like OurNakshatra, rather than attempting to calculate planetary degrees intrinsically.