Converts Hindi WordNet into 1.25M instruction pairs, fine-tunes a 12B model, and reports 91.0 pedagogical effectiveness for the resulting chatbot versus 79.4-83.6 for general models.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
From Lexicon to AI: A Structured-Data Pipeline for Specialized Conversational Systems in Low-Resource Languages
Converts Hindi WordNet into 1.25M instruction pairs, fine-tunes a 12B model, and reports 91.0 pedagogical effectiveness for the resulting chatbot versus 79.4-83.6 for general models.