OVA Odia Literature Dataset v1

About Dataset

This dataset is a curated monolingual corpus of Odia literary texts derived from books digitized by the Odia Virtual Academy (OVA). Each digitized book has been processed into a clean, UTF-8 encoded, machine-learning-ready CSV file in which every row represents a single sentence. Sentences are extracted using Odia-specific punctuation rules, primarily split by the Odia danda (।) and the question mark (?), ensuring linguistically consistent segmentation. By compiling diverse literary works into a uniform, sentence-level structure, this UTF-8–encoded dataset provides a robust foundation for tasks such as language modeling, translation, text classification, and broader computational studies of the Odia language.