We introduce Thovea, a THree-layer information diffusiOn Visual analytics systEm for lArge networks. This three-layered approach is designed to simultaneously investigate several diffusion processes---with each layer representing a different level of abstraction. Our method supports a top-down exploration strategy: starting from a high-level overview and drilling down to individual node details. This design provides a comprehensive understanding of the diffusion process(es) at hand, while enabling a detailed exploration down to distinct infection chains. We integrate suitable network layouts and representation methodologies into each level, aiming to support a scalable and agile exploration of information diffusion processes and networks. Information Diffusion investigates how information spreads over a tightly connected network and is relevant to many domains, such as modeling fake news spreading, pathogen contagion, or malware infections. While different modeling approaches exist in the literature, the visualization of information diffusion processes still constitutes an under-investigated problem. To gain a comprehensive overview, we present a survey and analysis of the current state-of-the-art in visual analytics techniques employed in representing and understanding diffusion processes happening over networks. We introduce a taxonomy that categorizes and structures the selected approaches while generalizing across application domains. Existing contributions are often tailored to niche domain-specific problems and lack generalizability. Visual scalability is also a challenge, as current research still struggles to effectively handle networks with thousands of nodes and edges, limiting their practical applicability. We evaluate our system by (i) presenting two case studies and (ii) conducting a quantitative value-driven estimation using the ICE-T methodology. The latter confirms the value of Thovea with a global average score of 5.82, where each component has been awarded a score greater than 5. The highest scored components are Essence and Insight with a score of 6.2 each.
dynamic networks
Analyzing large, dynamic network data, such as the Database of Modern Exhibitions, presents significant challenges due to the dataset's scale and complexity, as it tracks a decade of European art exhibitions and encompasses thousands of artists with evolving relationships. Centrality measures help address this complexity by using algorithms to quantify the importance of each node. However, an important next step is to explore how these calculated importance metrics can be transformed into meaningful visual representations to extract insights and identify key actors.To achieve this, we conducted a state-of-the-art literature review to investigate how centrality measures are applied in data visualization, which informed the development of dome-insights, a visual analytics tool that integrates centrality measures into both its visualization and interaction design. Dome-insights serves as a prototype to demonstrate how incorporating centrality measures can enhance the exploration of large, complex networks, facilitating insight discovery and identification of key actors.The tool was assessed by art historians and delivered positive results in both quantitative and qualitative evaluations, demonstrating its ability to uncover meaningful insights. More broadly, the findings highlight the potential of centrality measures in dynamic network analysis, underscoring their role in enhancing visual analytics for complex network data.