A complete, multi-layered quranic treebank dataset with hybrid syntactic annotations for classical arabic processing.

Area of Science:

• Computational Linguistics
• Natural Language Processing
• Digital Humanities

Background:

• Existing Quranic Treebank has limitations.
• Need for a comprehensive, multi-layered linguistic resource for Classical Arabic (CA).

Purpose of the Study:

• To describe the Extended Quranic Treebank (EQTB).
• To overcome limitations of the original Quranic Treebank.
• To provide a computationally accessible resource for CA.

Main Methods:

• Leveraged foundational data from established Quranic digital resources.
• Systematically expanded orthographic representations via algorithmic processing and validation.
• Refined morphological annotations using expanded schemas, automated re-annotation, and manual curation.
• Constructed a novel, complete syntactic layer using algorithmic conversion, Deep Learning-based parsing (hybrid constituency-dependency), and expert validation.

Main Results:

• Developed EQTB encompassing the entire Quran (~132,736 tokens).
• Structured data in adapted CoNLL-X format (43 columns).
• Detailed multiple orthographies, fine-grained morphology (45 tags), and complete hybrid syntax (140 tags/labels).
• Included auxiliary lexicons and schemas.

Conclusions:

• EQTB offers significant reuse potential for CA NLP tasks (parsing, morphology, diacritization).
• Provides crucial training/evaluation data for NLP.
• Supports linguistic research and development of pedagogical tools and language technologies.