[email protected] +91 9840521421 Scholarly Open Access, Peer-reviewed, Refereed Journal e-ISSN 2583-5300 Since 2022
  1. Home
  2. Archives
  3. 2026
  4. Vol. 5 Issue 3
  5. Article
Original Article

Development and Evaluation of a Conversational Chatbot for Maithili Using a Gemini API-Based Framework

Open Access Peer Reviewed Crossref DOI

Authors

  • Anu Priya1
  • Md. Irfan Alam2

1 Research Scholar, Jharkhand Rai University, Ranchi, Jharkhand, India. 2 Faculty of CSE & IT, Jharkhand Rai University, Ranchi, Jharkhand, India.

Article Information

Published Online
September-December 2026
Pages
953-958
Article Type
Original Article
Check for updates
check update

Abstract

View PDF

Conversational artificial intelligence has expanded access to information through natural-language interfaces, yet the quality of these systems can vary across languages with unequal digital representation. Maithili, an Indo-Aryan language with a substantial literary tradition, continues to require further language-specific computational resources and evaluation. This paper presents the development of a Python-based conversational chatbot that uses the Gemini application programming interface (API) to generate Hindi and Maithili responses to English-language queries. The application is designed to accept written and spoken input and to present responses in textual and audible forms. The work uses a Maithili linguistic resource collected by the Linguistic Data Consortium for Indian Languages (LDC-IL) as part of the research context. The paper describes the application workflow and proposes an evaluation framework focused on semantic adequacy, linguistic acceptability, fluency, contextual relevance, and recurring error types. Prior work has shown that automatic machine-translation metrics can correlate only moderately with human judgements for low-resource Indian languages, including Maithili, making human assessment particularly important. The available development draft establishes the system’s intended functionality and documents representative query categories; however, a reproducible quantitative evaluation has not yet been established in the materials available for this manuscript. Accordingly, no numerical performance claim is made. The study’s contribution is an application-level framework for examining API-based conversational generation for Maithili and a transparent evaluation protocol that can be completed using verified reference responses and human annotation.

APC from ₹1200 + 18% GST

Peer-reviewed open access₹1200 Indian (₹1400 with DOI)$80 USD internationalpaid after acceptance Peer-reviewed open access₹1200 Indian$80 USD international ₹1200 Indian · $80 intl

  • CrossRef DOI
  • ISSN 2583-5300
  • Open Access
  • Peer Reviewed
  • Plagiarism Checked
  • Google Scholar
Contact Us Submit Paper