Citation, License, and Colophon

Applied Natural Language Processing — From Tokens to Agents

Dr. Batuhan Bardak

Published by the author · First edition · Ankara, 2026

This version

This is a living book: chapters are revised as the field moves, and the Git history is the full audit trail. Cite a version, not “the website” — the prose you quote may have changed by the time your reader follows the link.

Edition record
Version v2026.09
Chapters in this version 39 of 39 and 6 appendices, in both English and Turkish
Language editions English · Türkçe
Home https://bbardakk.github.io/uygulamali-dogal-dil-isleme/
Source https://github.com/bbardakk/uygulamali-dogal-dil-isleme
DOI pending — see Persistent identifiers
ISBN not yet assigned

How to cite

Use the entry for the edition you actually read. The Turkish edition is a separate edition, not a translation appendix, and has its own entry.

@book{bardak2026applied,
  author    = {Bardak, Batuhan},
  title     = {Applied Natural Language Processing:
               From Tokens to Agents},
  year      = {2026},
  edition   = {v2026.09},
  publisher = {Published by the author},
  address   = {Ankara},
  url       = {https://bbardakk.github.io/uygulamali-dogal-dil-isleme/en/},
  urldate   = {2026-09-03}
}

APA 7:

Bardak, B. (2026). Applied natural language processing: From tokens to agents (Version v2026.09). https://bbardakk.github.io/uygulamali-dogal-dil-isleme/en/

To cite one chapter rather than the whole book, use @inbook with the chapter title and its stable URL — chapter URLs contain the chapter number, so they change if a chapter is renumbered. The version field is what makes your citation reproducible.

Include an access date. A living book has no single “published on” moment, so the access date is the only thing that tells your reader which text you saw.

Persistent identifiers

A DOI is being registered through Zenodo, which mints one identifier per released version plus a concept DOI that always resolves to the newest one. When it exists, this page and CITATION.cff will carry it, and it is the identifier you should prefer over the plain URL.

An ISBN has not been assigned. It identifies a published edition of a book rather than a work, so it belongs to the PDF and EPUB editions rather than to the website.

License

The text and the code are licensed separately, because they are used differently.

Licenses
Prose, figures, and exercises CC BY 4.0
Code, scripts, and configuration MIT

Under CC BY 4.0 you may share and adapt the text, including commercially, provided you give credit and indicate whether you made changes. Teaching from it, translating it, and building a course on it are all allowed and welcome; the citation above is the credit line.

Third-party material — datasets, model outputs, and quoted figures — stays under its own terms, which the chapter that uses it names at the point of use.

About the author

Batuhan Bardak, PhD, teaches the Applied NLP course at METU (ODTÜ) from which this book grew, to engineers outside computer science and to statistics students.

Colophon

Written in Quarto. Interactive figures are Observable JS and run in the reader’s browser — there is no server, and no data leaves the page. Static figures are hand-written SVG that follows the reader’s light or dark theme. Both editions build from one repository through GitHub Actions to GitHub Pages.

Every number printed in this book was produced by running the code, on the hardware named in each chapter’s environment table, and re-checked against the source before publication. Where a measurement failed to reproduce, or came out against the expected result, the book says so.

Disclaimer

The book is provided as-is, without warranty of any kind. Model behaviour, library APIs, and published benchmark numbers all change; verify anything you are about to depend on against your own data before you depend on it. That is, in fact, the argument of the book.