Transforming dance drama works into structured data through multi-dimensional computational analysis and combining it with audience emotional response assessment can evaluate dance drama works from a quantitative perspective, provide data support for the creation, performance and evaluation of dance dramas, and promote the innovative development of the dance drama art and the enhancement of audience experience. This study constructs a multi-dimensional computational analysis and audience emotional response assessment system for structured data of classical dance drama works. The study adopts a multidimensional structured data modeling framework to perform computational analysis of dance drama works from entity dimension, attribute dimension and history dimension, and extracts the core features of dance drama works by combining the multidimensional information fusion mechanism and the entity level encoder; at the same time, it designs an audience emotion assessment model based on ATAAE-BERT-BiLSTM to classify and identify the audience emotional response of dance drama works. The experimental results show that the proposed model outperforms the best-performing PIVOT in the baseline model by 2.88 and 0.67 in BLEU and NIST metrics, respectively; in the multidimensional correlation validation, the mean values of the structured data of the storyline, characterization, dance language, music performance, and stage scene reach more than 3000; in the emotion assessment model, the complete segmentation based on ATAAE-BERT-BiLSTM BERT-BiLSTM model achieved 83.25% accuracy and 83.88% F1 value in 4-segment segmentation mode, which were 5.14% and 6.72% higher than the BERT-LSTM model, respectively. The study provides a data-based method for the scientific evaluation of dance drama works and effectively supports the creation and dissemination of dance drama art.
