<p>Recommending visually coherent and stylistically diverse sets of artworks is a challenging task in digital curation, interior design, and personalized visual content services. Unlike traditional recommendation problems that focus on individual item relevance, composite artwork recommendation requires selecting a group of items that together satisfy a user’s stylistic intent while maintaining aesthetic compatibility. In this paper, we introduce a novel cloud-enabled graph-based framework for style-aware artwork composite recommendation. We construct an artwork correlation graph that models both the stylistic descriptors of individual artworks and their empirical compatibility based on historical co-occurrence. By leveraging distributed computation in cloud environments, our framework efficiently handles large-scale artwork collections and accelerates graph search. Given a user-defined set of style tags, our method identifies a minimal and connected subset of artworks that collectively cover the desired styles and form a coherent set in the graph. We formalize this task as a constrained subgraph selection problem and propose an efficient graph search algorithm supported by cloud-based parallelization to solve it. Experimental results on real-world artwork datasets demonstrate that our approach significantly outperforms existing baselines in terms of style coverage, visual harmony, and recommendation compactness. This work offers a principled and scalable cloud-oriented solution for generating aesthetically balanced and contextually appropriate artwork combinations.</p>