Modern public art is an important characteristic form of painting creation, which is constantly given new connotations with the development of the times and the progress of science and technology. Based on the characteristics of modern public art, the article explores the digitalization practice, creation process, and corresponding techniques for painting creation. The TF-IDF algorithm is introduced to mine the digital theme keywords expressed in painting creation, and the BERTopic model is combined to mine the digital theme of painting creation. The K-Means clustering algorithm is used for clustering the digital theme keywords of painting creation, and the digital theme keyword co-occurrence network is integrated to analyze the digital theme research trend of painting creation and expression for modern public art. When the number of digitized themes of painting creation and expression is 50, the F1 value of the BERTopic model can reach 94.36%, and the AUC area for digitized themes of painting creation and expression is 93.37%. The mean value of the keyword clustering profile coefficient of the digitized themes of painting creation and expression is 0.945, and the frequency of “painting creation” in the keyword co-occurrence network is up to 304 times. Relying on digital technology to create modern public art paintings, the digital theme expression of public art can be better realized.