RESEARCH ON THE APPLICATION OF TRANSFORMER DEEP LEARNING MODEL IN CONTENT ANALYSIS AND THEME MODELING OF THE PRAJNAPARAMITA HEART SUTRA

Abstract

This paper applies the Transformer model to topic modeling in Prajna Paramita Heart Sutra. Then, by pre-processing the original text and vernacular notes, it uses the Transformer model for theme modeling, optimizes the number of themes, optimizes the weight of keywords, and finally worked out five key topics: Prajna Paramita Heart Sutra, Buddhist theory, cultivation and practice, Buddha and Bodhisattva, Emptiness and Truth. These themes are quite consistent across various model outputs. These themes are categorized, keywords are extracted, and relationships are matched using the Transformer model at a quite high level of precision, recall, and accuracy. Meanwhile, this research confirms that, in the topic modeling process based on complex texts, the Transformer model is indeed effective and reliable, providing new technical support and a reference for studies in the field of classical Buddhist studies and text topic modeling. This is basic research work that can provide the basis for future investigation