Skip to content
PEBEducational Services
Dataset · TEXTBOOK

Textbook Dataset

Educational text carrying its ISBN, edition category, and page span, drawn from academic and institutional publishers across fifteen languages.

Textbooks
46K
Textbooks
Words
3B
Words
Languages
15
Languages

What it contains

The top entries by volume, with the long tail collapsed into a single figure.

Language

  • English1.3B45%
  • Arabic390M13%
  • Indonesian390M12%
  • Hindi310M10%
  • Telugu160M5%
  • Other (10 languages)400M13%

Record schema

Every field a delivered record carries, its type, and an example value.

FieldTypeDescriptionExample
isbnstring13-digit ISBN identifier, checksum-validated.9788169950817
titlestringTextbook title as printed..NET Application Development Lab
page_countintegerTotal pages in the source edition.161
word_countintegerExtracted word count for the full text.18345
categorystringSubject classification from the source catalog taxonomy.Computer Science & IT
stem_flagbooleanWhether the category is classified STEM or non-STEM.true
languagestringPrimary language of the extracted text.English

Scroll to see all columns

Sample records

Real rows, exactly as delivered. Copy them and check them.

Specimen · Textbook DatasetTextbook record
isbn9788169950817
title.NET Application Development Lab
word count18345
categoryComputer Science & IT
Verified
Textbook record · 2 shown
[  {    "isbn": "9788169950817",    "title": ".NET Application Development Lab",    "page_count": 161,    "word_count": 18345,    "category": "Computer Science & IT",    "stem_flag": true,    "language": "English"  },  {    "isbn": "9789379410634",    "title": "11th Maths Volume 1",    "page_count": 301,    "word_count": 121916,    "category": "Mathematics",    "stem_flag": true,    "language": "English"  }]

Provenance and rights

Sources, acquisition method, and the basis on which the material is held.

Source types
  • Academic and educational publishers
  • Institutional catalog partners
Acquisition
Direct acquisition from ISBN-attributed publisher and institutional catalog records.
Rights basis
Acquired under direct licensing agreements with publishers and institutional rights holders, tracked per ISBN.
Attribution fields
  • ISBN
  • Textbook title
  • Publisher category
License posture
Custom data license granted per engagement, scoped to pre-training and fine-tuning use.

Quality pipeline

Every record passes these stages in order. Each stage states what it removes.

  1. Ingestion and text extraction

    Extracts full text per page and reconciles the result against the printed page count.

    Files that fail extraction or show a page-count mismatch.

  2. ISBN validation

    Verifies the ISBN-13 checksum and confirms a catalog match.

    Invalid or non-existent ISBNs.

  3. Duplicate detection

    Compares ISBN and normalized title across the catalog.

    Exact duplicate records.

  4. Language identification

    Classifies language at the page level and reconciles against the declared language field.

    Records with unresolved language mismatches.

  5. Metadata completeness check

    Confirms title, category, and page count are present and non-null.

    Records with incomplete required fields.

Contamination screening

Named benchmarks checked for overlap, and the method used to check them.

Benchmarks checked
MMLU, GSM8K, MATH, ARC-Challenge, HellaSwag
Method
13-gram overlap screening against benchmark test and validation splits, consistent with decontamination practice used across large-scale pretraining corpora.

Delivery

Formats, access mechanisms, and how revisions are published.

Formats
  • JSONL
  • Parquet
Mechanisms
  • S3 pre-signed URL
  • Private Hugging Face repository
Versioning
Immutable dated snapshots, referenced by release tag.
Update cadence
Quarterly refresh.
Access model
Evaluation copy on request; full access under a signed data agreement.

Known limitations

What this dataset does not cover. Read this before you evaluate it.

  • Concentrated in English, Arabic, Indonesian, and Hindi; the remaining eleven languages each contribute a small share of total word volume.
  • Coverage reflects publisher and institutional catalog access at time of acquisition, not a complete survey of any language’s textbook output.
  • Category labels follow the source catalog’s subject taxonomy and are not independently re-verified against curriculum standards.
  • Text only - page images, figures, and diagrams are not extracted or included.

Request access to Textbook Dataset

We share sample records on request. Tell us your training stage and intended use and we will send an evaluation copy where the use fits.

Request sample records