Identifying prokaryotic and eukaryotic viruses is key to understanding host-environment interactions and ecological differences. Viral genomes, marked by rapid mutation and diversity, challenge accurate identification. We propose ViDSG, which hypothesizes that local nucleotide preferences and non-linear dependencies in viral sequences determine virus types. ViDSG quantifies nucleotide pair frequency and weighted distance to assess local combination preferences, while a masked language model captures global semantic features from non-linear dependencies. A dual-channel deep learning model integrates these statistical and semantic features for classification. We tested ViDSG using viral genome datasets from monkeys and pigs, comparing it with methods like IPEV and HTP for accuracy, robustness, and efficiency. ViDSG outperforms, especially in the 1200–1800 bp range, improving accuracy by 0.3%–49.4%. It maintains robustness with accuracy above 0.95 under 15% noise, surpassing competitors. Efficiency-wise, ViDSG is 4.1 times faster than IPEV, enabling rapid analysis of complex viral data with significant biomedical potential.