Using Machine Learning Technology to Decode the Bhagavad Gita | Technology
|Using Machine Learning Technology to Decode the Bhagavad Gita|
Decoding the Bhagavad Gita using Machine Learning Technology
This study paves the way for the application of AI-based tools to compare translations and assess sentiment across a variety of texts.
According to Eknath Easwaran, M.K. Gandhi and Purohit Swami's analysis of the quality of English translations of the Bhagavad Gita, machine learning and other artificial intelligence (AI) approaches have achieved enormous success in scientific and technological tasks such as determining how protein molecules are formed. and identify faces in a crowd. The use of these methodologies in the humanities, on the other hand, has yet to be substantially explored. But what can AI teach us about philosophy and religion? They used deep learning artificial intelligence algorithms to analyze English versions of the Bhagavad Gita, an ancient Hindu scripture initially written in Sanskrit, as a starting point for such research.
They investigated sentiment and semantics in translations using BERT, a deep learning language model. Despite considerable differences in language and sentence construction, they found that emotional and meaning-related patterns were largely comparable across all three. This study paves the way for the application of AI-based tools to compare translations and assess sentiment across a variety of texts.
About the Bhagavad Gita
The Bhagavad Gita is religious and intellectual literature in Hinduism. It was written nearly 2,000 years ago, has been translated into more than 100 languages, and has aroused the interest of Western scholars since the 18th century. The 700-verse composition is part of the larger Mahabharata epic, which describes the events of an ancient battle believed to have taken place near present-day Delhi in India at Kurukshetra. The text of the Bhagavad Gita recounts a discourse between Lord Krishna and a nobleman named Arjuna. They debate whether a soldier should go to war for ethical and moral reasons if he has close friends or relatives on the opposite side. Writing was instrumental in laying the foundations of Hinduism.
There have been numerous English translations of this holy book, however there is little evidence to support its excellence. Translations of poems and songs not only alter rhythm and rhyme structures, but can also lead to loss of semantic content. In 1785, the first of numerous English translations of the Bhagavad Gita was produced. In their study, they employed a deep learning language method to evaluate 3 selected translations of the Bhagavad Gita (from Sanskrit to English) using semantic and emotional analyses, which aid in the assessment of translation quality. They used BERT, a pretrained language model developed by Google. They further fine-tuned the algorithm using a human-labeled training dataset based on Twitter tweets that captures ten different attitudes. These sentiments (optimistic, grateful, empathic, pessimistic, anxious, sad, upset, denial, surprised, and joking) were adapted from their previous study of attitudes on social media during the onset of the COVID-19 pandemic.
The three translations they looked at had extremely varied vocabulary and structure, but the language model identified similar sentiments in different chapters of the translations. According to their model, the most frequently expressed emotions are optimism, irritation, and surprise. Furthermore, the model demonstrated how the overall polarity of feeling changes (from negative to positive) during the discourse of Arjuna and Lord Krishna. Arjuna is initially sad, but when Lord Krisha teaches him Hindu philosophy, he becomes hopeful. Krishna's sentiments demonstrate how, with philosophical knowledge of the dharma and mentorship, a troubled mind can find clarity to make the right decisions in times of conflict. His model's only restriction is that it was taught in Twitter data, so he identifies "jokes" as a popular sentiment. He inappropriately attaches this name to various passages in the Bhagavad Gita. Humor is nuanced and culturally bound, and understanding it would be asking too much of his model at this point. Different translators chose different words to explain the same principles considering the nature of Sanskrit literature; in reality, the Bhagavad Gita is a song with rhyme and rhythm, with the different dates of the versions.
The uses of sentiment analysis
His findings suggest the potential utility of AI-based tools for comparing translations and assessing sentiment in a variety of texts. This technique can also be used to analyze the emotions depicted in entertainment material. Another possible application is the evaluation of movies and songs to inform parents and officials about the suitability of the content for young people.