---
license: cc-by-4.0
task_categories:
- text-classification
language:
- ar
tags:
- Arabic
- stem
- education
- text-corpus
- nlp
- llm
- pretraining
- domain-specific
- educational-dataset
- scientific-text
pretty_name: Arabic STEM Educational Text Corpus Dataset
size_categories:
- 1B<n<10B
---
**Dataset Description:**

This dataset is a **large-scale collection of Arabic STEM Textbook data**, designed to support the development of advanced NLP systems and AI models for scientific understanding, analytical reasoning, and concept-based learning in Arabic.

The dataset includes structured and unstructured content from core STEM domains such as Physics, Chemistry, Mathematics, Biology, and Engineering, enabling models to interpret complex scientific concepts, solve problems, and generate accurate, context-aware responses in Arabic.
Additionally, this dataset can be used in pipelines for **Supervised Fine-Tuning (SFT) and Reinforcement Learning with Human Feedback (RLHF) workflows**, improving model performance in technical comprehension, reasoning, and problem-solving tasks.

**Key Use Cases**

    -Text-based Question Answering 
    -Named Entity Recognition (NER) in STEM content
    -Scientific text summarization and comprehension
    -Automated tutoring and educational assistants
    -STEM knowledge retrieval systems
    -Model evaluation and benchmarking

**Dataset Specification**

    -Modality: Arabic
    -Type: Educational / STEM
    -Data Source: Curated academic textbooks and educational material
    -Data Nature: Real-world and curated data
    -Content: Scientific theories, formulas, numerical problems, explanations, and conceptual text
    -Books: 1,238

**Value of This Dataset**

    -Enables learning of STEM subjects in Arabic
    -Improves analytical reasoning and problem-solving in AI models
    -Supports multilingual and domain-specific NLP systems
    -Helps build AI-powered educational platforms
    -Enhances accuracy and reliability of LLMs in STEM domains

**Basic JSON Schema**
```json
{
  "pdf_name": "string",
  "pages": [
    {
      "content": "string",
      "page_number": "integer"
    }
  ]
}
```
**Full Dataset Overview**

  2.28+ Billion Words covering 5,000+ Subjects with interwoven images for deeper context 32,000+ Books in 14 Languages

**Data Creation**

  Procured through formal agreements and generated in the ordinary course of business.

**Considerations**

This dataset is provided for research and educational purposes only. It contains only sample data. For access to the full dataset and enterprise licensing options, please visit our website [InfoBay AI](https://infobay.ai/) or contact us directly.

    -Ph: (91) 8303174762
    -Email: vipul@infobay.ai