Zdteks

zdteks is a lightweight, open‑source toolkit designed for extracting structured data from unstructured textual documents. Developed in 2022 by a consortium of research laboratories at the University of Grenoble and the Institute of Data Science in São Paulo, the project aims to provide a unified interface for natural language processing, regular expression matching, and machine learning‑based entity recognition. The name “zdteks” derives from the French phrase “Zéro Dépendance–Texte,” reflecting the library’s emphasis on minimal external dependencies while working primarily on text.

The core of zdteks revolves around a modular architecture that allows users to chain processing steps—tokenization, segmentation, named entity extraction, and custom rule application—into pipelines defined through simple configuration files. The library supports multiple programming languages; primary bindings exist for Python, Java, and Rust, and can integrate with popular data processing frameworks such as Apache Spark and Pandas.

In practice, zdteks is employed in domains that require rapid prototyping of text analytics workflows, including digital humanities, legal document review, and medical record summarization. Its lightweight footprint makes it suitable for deployment on edge devices and serverless environments, and the open‑source license (MIT) encourages community contributions. Since its first public release, the project has accumulated over 2,000 stars on GitHub and has been cited in several academic papers on information extraction.

Future development plans for zdteks include support for multilingual processing, incorporation of transformer‑based models for higher‑accuracy entity recognition, and a more extensive set of pre‑built pipelines for domain‑specific applications. The project’s governance is overseen by the Zdteks Foundation, which organizes annual code sprints and maintains an online forum for user support.