Najm Institute (موسسه نجم)

Pioneers in Semantic Intelligence: Applying Explainable AI to Islamic Jurisprudence.

Najm Institute is a research organization founded in 2018 with the core mission of applying the frontiers of Explainable Artificial Intelligence (XAI) to the process of ijtihad (the independent legal reasoning of a jurist in Islamic law).

We bridge the gap between cutting-edge AI and traditional Islamic sciences, developing advanced computational tools, large-scale datasets, and intelligent systems. Our work is built on a strong academic foundation in collaboration with Iran University of Science and Technology and is organized into two main pillars: industrial products and foundational research.

• 🌐 Official Website: najm.ac
• ✉️ Contact: info@najm.com

🚀 Our Projects & Solutions

We have successfully developed several high-maturity products and datasets that are actively in use:

📂 Rezwan Corpus: An Ocean of Hadith Data

The Rezwan Corpus is the most comprehensive intelligent database of Shia and Sunni hadith, containing over 1.3 million narrations from authentic sources. Enriched with the power of AI, this dataset is designed to empower researchers, academics, and developers by providing structured access to one of the largest collections of hadith in the world. The corpus is available here on the Hugging Face Hub.

• Access the Rezwan Corpus Dataset

Thiqat: Intelligent Hadith Knowledge Graph (TRL 8)

Our flagship product, Thiqat, is a comprehensive knowledge graph of Shia hadith. It integrates:

• All Shia hadiths and their chains of transmission (sanad).
• Biographies and reliability data for all narrators (rijal).
• A complete catalog of hadithic books.

The platform provides services for intelligent reasoning (machine thinking), ontology-based knowledge retrieval, and semantic search and chatbots for hadith and narrator data.

Johd: Intelligent Search for Islamic Sciences (TRL 8)

Johd.ir is a public semantic search engine covering tens of thousands of Islamic books. Its capabilities include intelligent entity recognition, identification of proper nouns, and discovery of related search queries.

Semantic Search for Islamic Rulings (TRL 8)

This question-answering system (qa.ahdsoft.dev) utilizes vector search technology to provide intelligent access to a database of over one million questions and answers on Islamic rulings (fiqh).

Other AI Services (TRL 6-7)

We have also developed a suite of specialized NLP tools, including:

• Mihan: An intelligent entity recognition system.
• Sahab: A search engine for text within image-based (scanned) books.
• Services for keyphrase extraction, corrupted text detection, and intelligent text generation.

🔬 Our Research Areas

Our products are backed by dedicated research groups that continuously push the boundaries of AI and its application to religious texts. Our primary research areas include:

• Knowledge Graph Processing: Modeling and reasoning over large-scale, interconnected scholarly data.
• Machine Reading & NLU: Deepening the machine's understanding of the nuances of classical Arabic and Persian.
• Logic of Reading Religious Texts: Formalizing the logical frameworks used in textual interpretation.
• Intelligent Question Answering: Developing systems that can understand and answer complex user queries.
• Commonsense Reasoning: Imbuing AI models with the background knowledge required to understand context.
• Computational Argument Mining: Automatically identifying and structuring arguments within scholarly debates.
• Machine Thinking: A newly established group (2024) exploring advanced reasoning, with supervision from professors at the University of Luxembourg and Sohar University, Oman.