Progressive Visual Analytics for Network Traffic Analysis

Summarise with VIRAT
Authors: Hannah Klohr
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Visualizations are essential for data analysis. However, large, complex data sets, such as network traffic data, can cause latency, cognitive load, and slow interactions. Progressive visualization offers a promising solution by providing successive, approximate visual updates that enable early feedback and continuous exploration. Guided by the goal of understanding how progressive visualization affects user confidence, accuracy, and decision making during network traffic analysis, we extended the network traffic analysis platform NetCapVis with progressive functionality, including a progressive interface, progressive data loading and buffering, and an interactive steering module. We reviewed foundations in information visualization, progressive analytics, and network traffic analysis, as well as related evaluation methodologies. Based on this, we developed a conceptual model for progressiveness in NetCapVis, defined design requirements, and implemented buffering with a progressive packet queue, decoupled loading and display algorithms, and a progression panel for controlling update speed and navigation. Random sampling was used as the primary sampling method. The resulting system met all defined requirements and enabled perceptually stable, informative, and interactive progressive updates. Further, we contribute a controlled user study assessing how confidence, correctness, and decision making evolve during progression. Using a simplified pattern-recognition task, participants viewed progressively revealed visualizations in two blocks, either rating confidence at fixed progression steps or choosing when to stop. Results showed a strong increase in confidence with progression, high correctness after limited data exposure, and clear differences between pattern types. Participants stopped early (on average at 24.72%) while maintaining high accuracy. All hypotheses regarding confidence growth, decision accuracy, pattern effects, and early stopping were accepted. Qualitative feedback further highlighted the perceived usefulness and potential of progressive visualization to improve interactivity and analysis speed. Overall, the thesis demonstrates both the feasibility of a progressive prototype in NetCapVis and the measurable benefits of progression on user performance.

Fields of Study & Subjects

Data Visualization and Analytics Computer Graphics and Visualization Techniques Visual and Cognitive Learning Processes

Publication details

Journal / source
Fraunhofer-Publica (Fraunhofer-Gesellschaft)
Publication date
1 January 2026
Language
EN
Document type
Thesis