RACTrix: Interactive Visual Summaries for Exploring Signed Networks (Poster Abstract)

Abstract

Force-directed network layouts naturally reveal densely connected subgraphs or communities as distinct spatial clusters. However, since node positions are used to encode the graph structure, other node and edge attributes must be represented through secondary visual encodings such as color, shape, or edge thickness. This makes it difficult to accurately interpret quantitative information such as node rankings, community sizes, and edge properties without additional interactions, such as hovering, clicking, or issuing queries. The challenges become even more severe for signed graphs, where edges can be weighted and carry either positive or negative labels. Here we present RACTrix, a compact, intuitive, and interactive visual summary to aid in the exploration of signed networks. RACTrix combines the aesthetics of right-angle crossing (RAC) drawings with a matrix-like representation and links them with a force-directed layout. RACTrix provides positional encoding for node/cluster rankings, edge rankings around a node/cluster, edge signs, edge directions, and cluster sizes, while it maintains a matrix-based representation of pairwise relationships among nodes/clusters. We coordinate RACTrix and force-directed layout such that selecting an element in one view highlights the corresponding elements in the other. This allows for exploring quantitative network features alongside the network structure which is crucial for validating visual patterns in the network layout.