About anvay

A tool for exploring thematic structure in text

The word anvay (অন্বয়) carries layered meanings in Bengali: connection, syntax, succession, logical consequence, lineage. This project takes its name from that rich semantic field — the idea of tracing meaningful relationships through language.

anvay is an open-source application for topic modelling and visual interpretation of text corpora. It is designed to support digital humanities research, literary analysis, and linguistic inquiry across multiple languages. Built around Latent Dirichlet Allocation (LDA), anvay integrates language-specific preprocessing workflows, structured visualisations, and relevance scoring to enhance the interpretability of computational topics.

Unlike general-purpose topic modelling tools, anvay has been built with the specific challenges of South Asian languages in mind — morphological richness, script complexity, and the relative scarcity of NLP resources compared to English. It enables scholars, students, and researchers to upload their own textual materials and receive a structured breakdown of latent themes, token frequency distributions, and interpretive clues.

Language Support

anvay currently supports four languages. Bengali is the primary supported language, with the most mature and extensively tested preprocessing pipeline. English, Hindi, and Tamil are experimental — functional, but less validated, and we welcome feedback from researchers working in these languages.

Institutional Support

anvay is a project of DHLAB, a space for computational and interpretive experiments at the intersection of humanistic inquiry and digital methods. DHLAB is housed in the School of Humanities and Social Sciences at Shiv Nadar Institution of Eminence, and is one of the first dedicated digital humanities laboratories in India. Its primary focus is on South Asian languages and archives.

The development of anvay has been supported by the School of Humanities and Social Sciences at Shiv Nadar Institution of Eminence.

Citing anvay

If you use anvay in a scholarly publication, please cite:

Das Gupta, V., (2026). anvay: A Web-based Tool for Interpretive Topic Modelling in Bengali. Journal of Open Source Software, 11(118), 8641, https://doi.org/10.21105/joss.08641

@article{Das Gupta2026,
  doi       = {10.21105/joss.08641},
  url       = {https://doi.org/10.21105/joss.08641},
  year      = {2026},
  publisher = {The Open Journal},
  volume    = {11},
  number    = {118},
  pages     = {8641},
  author    = {Das Gupta, Vinayak},
  title     = {anvay: A Web-based Tool for Interpretive Topic Modelling in Bengali},
  journal   = {Journal of Open Source Software}
}