Jaipur, Nov. 4 -- Researchers at Indian Institute of Technology (IIT) Jodhpur are developing an artificial intelligence (AI)-powered platform to decipher, restore and digitise ancient Indian manuscripts, potentially opening up centuries-old knowledge that has remained inaccessible due to fading ink, damaged pages and complex historical scripts, states an official release from IIT. The research initiative, led by Dr Dinesh Mohan Joshi, assistant professor at IIT Jodhpur's School of Liberal Arts, seeks to combine AI, computational linguistics and Indian Knowledge Systems to make India's manuscript heritage more accessible to researchers, historians, linguists and scholars. The proposed platform will employ an adaptive AI framework capable of identifying the script and language of a manuscript and selecting appropriate processing techniques depending on its physical condition and textual characteristics. It will integrate multiple technologies, including image restoration, page-layout analysis, text segmentation and handwritten text recognition, to convert manuscript images into machine-readable digital text, it states. Unlike conventional optical character recognition (OCR), which is primarily designed to extract text from printed documents, the proposed system is being developed to address the specific challenges posed by historical handwritten manuscripts. These include faded and smudged writing, damaged or incomplete characters, complex ligatures, marginal notes and variations in handwriting across different periods and scribes. The platform is designed to support six Indian scripts - Devanagari, Sharada, Grantha, Nandinagari, Modi and Tulu-Tigalari - with provisions to process classical and regional Indian languages. It will also incorporate language-processing tools for transliteration, text normalisation and translation into modern Indian languages and English, enabling scholars unfamiliar with historical scripts to access contents of manuscripts. An important feature of the proposed system will be its ability to indicate the reliability of its interpretations. The digitised text will carry confidence scores, allowing researchers to identify uncertain readings and verify them against the original manuscripts. In cases where automated recognition is inadequate, the framework will allow experts to intervene and correct the interpretation, ensuring that the original meaning and authenticity of the material are not compromised. The research also draws on Dr Joshi's academic work involving unpublished manuscripts related to India's mathematical and astronomical traditions. These include a commentary on Brahmagupta's Brahmasphutasiddhanta by the 10th-century scholar Prithudakasvami and the Suryasiddhanta, accompanied by a commentary by 16th-century scholar Vishvanatha. Such manuscripts contain valuable historical knowledge, but accessing and interpreting them can be difficult because of their age, condition and specialised scripts. The proposed AI framework could help overcome these barriers by providing a systematic approach to recovering, transcribing and analysing handwritten historical texts. The initiative is aligned with the Centre's Gyan Bharatam Mission....