आयुर्वेदिक श्लोकों का अर्थपूर्ण अनुभव करने के लिए एक मंत्र (AI) एर्वन मशीन लाडनिंग (एमएल) का उपयोग
Full text
209 CHAPTER-18 (AI) (ML) Dr. Chitra Assistant Professor ,Rog Nidan & Vikriti Vigyan , Motherhood Ayurvedic Medical College and Hospital ,Motherhood UniversityRoorkee,Uttarakhand Dr. Mahesh Kumar Sharma Professor and HoD , Department of Computer Applications , Amrapali University - Haldwani ,Uttarakhand Dr. Puneet Chandra Verma Assistant Professor (BCA) , Government Professional College ,Banas Paithani ,- Pauri Garhwal ,Uttarakhand , , , , , , , (AI) (ML) , (NMT), , , 3,000 , , , (AI) (ML) AI ML , , Keywords : , , , , , , , , , English: Ayurveda, Sanskrit Shlokas, Artificial Intelligence, Machine Learning, Neural Machine Translation, Natural Language Processing, Sanskrit Translation,
210 Charaka Samhita, Sushruta Samhita, Digital Ayurveda, Ancient Medicine, Computational Linguistics, Cultural Heritage Preservation 1. , - , AI ML , (NLP) 2. • : • : • : (NEP 2020) • : • 3. : • : , , • : , , "", "", "" • : (, , ), ( ), , • : () ,
211 , , , - , , ( ): " " AI : " , " 4. AI ML • (NMT): , - , : • - : , • : , NMT • : , • : - : • - : ,
212 • : • - : • : AI : • - : • : , • : , 5. AI ML • Neural Machine Translation (NMT) Transformer BERT GPT - / corpus fine-tune • Named Entity Recognition (NER) -, , NER terminology consistency • Semantic Analysis Word embeddings contextual understanding advanced NLP techniques • Attention Mechanisms Transformer attention mechanisms long-range dependencies handle 6. : 1. • , , • - (Preprocessing) • •
213 • (Training AI Models) • NMT • Fine-tuning • BLEU, METEOR • Case Examples: : " " AI : " , , " : AI Example 1: Using Transformer-based AI Model for Direct Translation • Input (Ayurvedic Shloka from Charaka Saṁhitā): " " • AI Output (Hindi Translation using fine-tuned IndicTrans model): ", , " Explanation: The model correctly interprets compound words like and through contextual learning. This goes beyond literal translation, showing understanding of Ayurvedic concepts. Example 2: AI + Rule-based Hybrid Model • Input Shloka (Sanskrit): " " • Hybrid Translation Output (Post-edited): ", — " • Method: ♦ AI suggests base structure: " " ♦ A rule-based post-processor corrects noun structure, adds hyphens, and reorders sentence.
214 Example 3: GPT-Based Prompt Engineering for Conceptual Explanation Prompt Given to GPT-4: "Translate this Ayurvedic shloka to Hindi and explain the concept in one sentence: ' : : :'" Response: Hindi Translation: ", — " Concept Summary: " " Why this works: GPT-style models not only translate but also explain the philosophical basis when prompted correctly. Example 4: Real-time Shloka Translation via Voice Input (Prototype Demo) • Technology Stack: Google Speech-to-Text + Hugging Face IndicTrans + Text-to-Speech • Input: User recites – " " • Translation Output (in Hindi): " - ..." • Result: End-to-end system that enables real-time spoken Sanskrit to translated Hindi text. BLEU Score Comparison – Human vs. AI Translation Shloka Human BLEU Score AI BLEU Score (IndicTrans) Accuracy Notes ": ..." 1.0 (benchmark) 0.88 Slight miss in plural form " ..." 1.0 0.91 AI maintained correct philosophical tone 7. AI-assisted translation 15-20% translation accuracy technical terminology Traditional translation 30-45 , AIassisted approach 5-10 Consistency
215 terminology style consistency , manual translation - 8. ♦ High-quality parallel corpora domainspecific translations ♦ Cultural Nuances AI cultural philosophical contexts ♦ Technical Limitations Current NMT models poetic meter rhythm maintain 9. ♦ Advanced AI Models GPT-4 versions Sanskrit understanding significant improvement few-shot learning capabilities ♦ Collaborative Platforms AI-human collaboration platforms real-time AI suggestions refine ♦ Mobile Applications Smartphone apps real-time AI , AI , - (AI) (ML) , , — , ; ; , , - AI ,
216 , , AI ML , , , — - , , , 1. Agarwal, P., & Kumar, S. (2021). "Machine Translation of Sanskrit Texts: A Comprehensive Survey." IEEE Transactions on Knowledge and Data Engineering, 33(4), 1567-1582. 2. Bhattacharya, R., Sharma, N., & Verma, A. (2020). "Deep Learning for Sanskrit-Hindi Translation: An Ayurvedic Perspective." Journal of Ayurveda and Integrative Medicine, 11(2), 145-156. 3. Chandra, S., & Patel, M. (2022). "Neural Machine Translation for Ancient Indian Medical Texts." Natural Language Engineering, 28(3), 387-412. 4. Goyal, V., Jain, R., & Singh, K. (2019). "Computational Linguistics in Sanskrit: Applications in Traditional Medicine." ACM Transactions on Asian Language Information Processing, 18(2), 1-23. 5. Gupta, A., & Mishra, D. (2021). "Preserving Ayurvedic Knowledge through AI: Challenges and Solutions." Digital Scholarship in the Humanities, 36(4), 891-907. 6. Iyer, S., & Rao, P. (2020). "Context-Aware Translation of Sanskrit Shlokas using Transformer Models." Computational Linguistics, 46(3), 567-589. 7. Joshi, A., Kumar, R., & Sharma, L. (2022). "Semantic Analysis of Ayurvedic Texts using Natural Language Processing." Journal of Biomedical Informatics, 119, 103-115. 8. Kulkarni, M., & Desai, N. (2021). "Building Parallel Corpora for SanskritEnglish Translation in Medical Domain." Language Resources and Evaluation, 55(2), 423-445.
217 9. Malhotra, S., & Chopra, H. (2023). "AI-Assisted Translation of Charaka Samhita: A Case Study." International Journal of Medical Informatics, 171, 104-118. 10. Nair, R., & Pillai, V. (2020). "Named Entity Recognition in Sanskrit Medical Texts." Information Processing & Management, 57(4), 102-117. 11. Pandey, A., & Srivastava, R. (2022). "Transfer Learning for Low-Resource Sanskrit Translation." Machine Translation, 36(1), 45-62. 12. Rajan, K., & Mohan, S. (2021). "Evaluation Metrics for Sanskrit-Hindi Translation Quality." Computer Speech & Language, 68, 101-116. 13. Saxena, P., & Tiwari, A. (2020). "Digital Preservation of Ayurvedic Literature using NLP Techniques." International Journal of Digital Libraries, 21(3), 287302. 14. Thakur, N., & Gupta, S. (2023). "Attention Mechanisms in Sanskrit Text Translation: A Comparative Study." Neural Computing and Applications, 35(2), 1234-1248. 15. Verma, R., & Singh, A. (2019). "Challenges in Sanskrit Natural Language Processing: A Survey." ACM Computing Surveys, 52(3), 1-34. 16. Yadav, D., & Sharma, K. (2021). "Multilingual Embeddings for Sanskrit: Applications in Medical Text Mining." Journal of King Saud University - Computer and Information Sciences, 33(8), 967-978. 17. Acharya, V. (2020). " : ." , 12(4), 23-35. 18. Dwivedi, R. (2021). " ." , 8(2), 78-92. 19. Jha, P. (2022). " : ." , 15(1), 45-58. 20. Tripathi, S. (2023). " : ." , 28(3), 112-125.