Visualizing software graphs efficiently is crucial for understanding complex software architectures. Traditional techniques struggle with large, sparse, heterogeneous graphs containing semantically diverse relationships. This paper presents SemViSG, an approach that combines semantic-enriched hierarchical community detection, and custom layout algorithms to enhance software graph comprehension. We introduce multi-layered software graph representation and semantic enrichment techniques that leverage source code semantics to create functionally coherent communities. The custom layout algorithm preserves directional flows while minimizing edge crossings. Validation on large-scale systems demonstrates substantial improvements in clarity and usability, providing software engineers with enhanced architectural insights.