Abstract
Abstract This study proposes a digital economic development evaluation algorithm that integrates knowledge graphs and contrastive learning. By constructing a digital economic knowledge graph containing 120,000 entities and 650,000 relationships, the representation of entity relationships is optimized by combining the graph contrastive learning module. The experiment uses the self-built DigitalEconomy-2024 dataset. Verification shows that KGC-DEE outperforms the baseline models on core indicators, achieving an accuracy of 0.923 and an F1-Score of 0.918, while the root mean square error and average absolute error are reduced to 0.068 and 0.052, respectively. The research confirms the synergistic advantages of knowledge graph-structured prior knowledge and Contrastive Learning-based discriminant representation. It provides an innovative tool for quantitative evaluation of the digital economy.