Abstract:
ChatGPT is a large-scale neural language model which is developed by OpenAI. This belongs to the family of Generative Pretrained Transformers (GPT) models, which are based on a deep neural network architecture known as Transformers. According to the researchers ChatGPT has been trained on a massive corpus of textual data using unsupervised learning, which means that it has learned to understand the patterns and relationships between words and phrases in natural language without being explicitly taught. ChatGPT has shown promising results in literary text applications such as translation, summarization, and the generation of new texts. In literary translation, ChatGPT can be trained on a corpus of texts in different languages and can generate high-quality translations into the target language. This is particularly useful in situations where human translators are not readily available, or the budget of the professional translation is limited. Hitherto most researchers have identified other machine translation tools like Google translate to be ineffective when compared to manual translations in literary contexts. Thus, this research aims to analyze the effectiveness of ChatGPT in translating selected Sinhala songs into English language, using quantitative performance metrics known as “Precision” and “Recall” in the Natural Language Processing Model (NLP). Moreover, this will also evaluate the accuracy of ChatGPT in generating poems on a given subject. In addition, a qualitative inductive research approach will also be used by researchers to contextually analyze context related to the cultural sociological effectiveness of poems generated by this software. This research will be helpful in situations where human translators are not readily available, or translation cost is very high.
