Assessment of Digital Twin Virtual-Real Connected Interaction Based on NLP

Digital twin is a key technology to realize the interactive integration of the physical world and the virtual world, showing many application values in intelligent operation and maintenance, virtual debugging, anomaly diagnosis, risk prediction, decision-making assistance, system optimization and other scenarios. Artificial intelligence methods such as machine learning and natural language processing not only drive the development of digital economy, but also disrupt the research paradigm of technology evaluation due to their robustness and efficiency. This paper aims to apply natural language processing processes and machine learning algorithms to achieve the assessment and visualization analysis of the degree of interaction of digital twins with virtual and real connections based on multi-label text classification tasks. We collected a novel text dataset consisting of Chinese digital twin application cases in 2 decade, and based on the theory of connected interaction, trained four machine learning models including SVM, DT, RF and Xgboost to complete the classification task, and visualized the relationship between the degree of Chinese digital twin applications real-virtual interactions, city clusters, and industries. Experimental results show that Xgboost, an integrated tree-based machine learning algorithm, has a stronger ability to learn from domain experts, which is beneficial to mitigate expert bias in technology assessment and reduce organizational costs, and provides practical evidence.