AggreSet: Rich and Scalable Set Exploration using Visualizations of Element Aggregations

Area of Science:

• Information Visualization
• Data Mining
• Human-Computer Interaction

Background:

• Datasets frequently contain multi-value (set-typed) attributes, leading to complex relationships.
• Existing exploratory tools for set-typed data lack scalability and interface consistency.
• Combinatorial growth of relations poses challenges for analyzing numerous sets.

Purpose of the Study:

• Introduce AggreSet, a novel system for scalable and contextual exploration of set-typed data.
• Improve the consistency and efficiency of exploratory tasks like selection and comparison for set-typed attributes.
• Address the scalability challenges associated with increasing numbers of sets and their intersections.

Main Methods:

• AggreSet generates aggregations for sets, set-degrees, set-pair intersections, and other attributes.
• Utilizes matrix plots for set-pair intersections and histograms for set lists and degrees.
• Implements a non-overlapping visual design scalable to numerous and large sets.

Main Results:

• AggreSet effectively visualizes element counts per aggregate, enabling scalable analysis.
• The system supports core exploratory tasks including selection, filtering, and comparison.
• Demonstrated effectiveness across diverse datasets and validated through expert reviews and a case study.

Conclusions:

• AggreSet provides a scalable and consistent framework for exploring complex set-typed data.
• Facilitates rich analysis of set relations, including subsets, disjoint sets, and intersection strength.
• Offers a powerful tool for data exploration and pattern detection in set-based datasets.