Vedalytics: Concepts as Knowledge from Rules of the Ancient Indian Scriptures
Animesh Chaturvedi, V K Vivekraj, Pratik Pakhale, Karthik Avinash · 2024
Bharat (India) is an ancient land where oldest Sanskrit literatures are discovered including Vedas, Puranas, Upanishads, etc., collectively these sacred scriptures (i.e, Ved) can be called as Ancient Indian Scriptures (AIS). Understanding vast Sanskrit literature (big textual data) is a laborious task for a group of persons during their life-time. In this work, we have curated several AIS into a repository. We propose a Vedalytics (Ved Analytics) approach to retrieve Scripture Rules and concepts from ancient-words in AIS. This provides insightful knowledge about the content of each AIS. We have analysed translations of verses (Shlokas) of 4 Vedas (Rigveda, Yajurveda, Samaveda, and Atharvaveda); Shlokas of Bhagavad Gita and Ramayana; and translations of Upanishad; translations of 3 Puranas; 7 different compositions and commentaries on AIS. We retrieved rules on these multi-lingual AIS (Sanskrit, Hindi, and English) for knowledge discovery such as: a) interesting concepts and b) transitivities and lattices relationship between ancient-words. We present metrics for Coverage, Information Compression, and Knowledge Discovery to measure the efficiency of summarization. We found that our approach generates specific and contextual results as compared to generic LLMs. Repository link: https://github.com/animesh88/AncientTextAnalytics