Semantic and sentiment analysis of selected Bhagavad Gita translations using BERT-based language framework

  • 2022-01-09 23:59:11
  • Rohitash Chandra, Venkatesh Kulkarni
  • 2

Abstract

It is well known that translations of songs and poems not only breaks rhythmand rhyming patterns, but also results in loss of semantic information. TheBhagavad Gita is an ancient Hindu philosophical text originally written inSanskrit that features a conversation between Lord Krishna and Arjuna prior tothe Mahabharata war. The Bhagavad Gita is also one of the key sacred texts inHinduism and known as the forefront of the Vedic corpus of Hinduism. In thelast two centuries, there has been a lot of interest in Hindu philosophy bywestern scholars and hence the Bhagavad Gita has been translated in a number oflanguages. However, there is not much work that validates the quality of theEnglish translations. Recent progress of language models powered by deeplearning has enabled not only translations but better understanding of languageand texts with semantic and sentiment analysis. Our work is motivated by therecent progress of language models powered by deep learning methods. In thispaper, we compare selected translations (mostly from Sanskrit to English) ofthe Bhagavad Gita using semantic and sentiment analyses. We use hand-labelledsentiment dataset for tuning state-of-art deep learning-based language modelknown as \textit{bidirectional encoder representations from transformers}(BERT). We use novel sentence embedding models to provide semantic analysis forselected chapters and verses across translations. Finally, we use theaforementioned models for sentiment and semantic analyses and providevisualisation of results. Our results show that although the style andvocabulary in the respective Bhagavad Gita translations vary widely, thesentiment analysis and semantic similarity shows that the message conveyed aremostly similar across the translations.

 

Quick Read (beta)

loading the full paper ...