AI-assisted reading and digital infrastructure for Vedic philology
VEDA AI is an AI-assisted reading environment for Vedic philology, developed together with a set of digital resources for working across electronic texts, printed editions, page images, dictionaries, linguistic annotation, and scholarship.
The project
Vedic research depends on moving across electronic texts, historical editions, scans, dictionaries, grammatical annotation, and more than a century of scholarship. VEDA AI introduces computational assistance into that reading process while keeping the underlying texts and sources visible.
The project combines the AI reading environment itself with open digital resources that support comparison, verification, discovery, and reuse of philological evidence.
Research
AI-assisted reading
An integrated environment for reading Vedic texts with computational support for morphology, lexical lookup, correction, annotation, OCR, and related analysis.
Text & Image
Connecting printed editions and page images with machine-readable texts so that philological evidence remains visible and verifiable.
Language & Knowledge
Linking linguistic annotation, dictionaries, corpora, bibliography, and secondary scholarship into reusable research infrastructure.
Research direction
VEDA AI & Vedic resources
Core project
An environment for both reading with AI (AI-assisted reading) and AI reading autonomously.
↗Open
A parallel viewer for Vedic electronic texts, printed editions, and page images.
↗Open
A collection that brings together IIIF-accessible page images of primary and secondary sources used in Vedic research.
↗Open
A web edition of the JAIBS bibliography and its abbreviations for Vedic studies.
↗Open
A search interface for abbreviations used in EWAia.
↗Outputs
Selected publications directly connected to the VEDA AI research programme are listed here.
Machine translation for Vedic Sanskrit, supported by a parallel corpus aligning Vedic Saṁhitās with existing translations.
Machine translation augmented with philological commentary, connecting translation output with scholarly interpretive context.
OCR for printed Vedic Sanskrit, including recognition of Vedic accent marks.
Automatic restoration of missing Vedic pitch accents with neural language models.
Project lead
VEDA AI is led by Yuzuki Tsukagoshi, whose work spans Vedic linguistics and philology, digital humanities, and natural language processing. Research profile
Funding
VEDA AI and its related research infrastructure are supported by competitive research funding from JST and JSPS.
JST BOOST
This project develops machine translation and annotation-generation models for Vedic Sanskrit. It combines language models with external scholarly knowledge, preserves Vedic-specific textual features, and develops a platform that incorporates expert human evaluation into annotation generation.
JSPS KAKENHI
A Grant-in-Aid for Early-Career Scientists project (FY2025–2029) connecting electronic texts, printed images, and primary and secondary Vedic literature through OCR and multilingual machine-learning models.
JSPS KAKENHI
A Grant-in-Aid for Transformative Research Areas (Research in a proposed research area) project (FY2026–2028) developing methods to extract and analyse entities specific to historical materials and infer relationships among documents using language models.