MatLink: Enhanced Matrix Visualization for Analyzing Social Networks

Abstract

Abstract. Visualizing social networks presents challeges for both nodelink and adjacency matrix representations. Social networks are locally dense, which makes node-link displays unreadable. Yet, main analysis tasks require following paths, which is difficult on matrices. This article presents MatLink, a hybrid representation with links overlaid on the borders of a matrix and dynamic topological feedback as the pointer moves. We evaluated MatLink by an experiment comparing its readability, in term of errors and time, for social network-related tasks to the other conventional representations on graphs varying in size (small and medium) and density. It showed significant advantages for most tasks, especially path-related ones where standard matrices are weak.

Visually Exploring Large Social Networks

“…The second experiment I performed was much more focused, using five tasks and a technique I developped to generate representative datasets 4 . It ended with significant results showing that MatLink improved matrices [16]. However, it is hard to generalized these results to the global process of exploration.…”

“…MatLink: Improving Matrix Representations [16]. Ordering a matrix helps identifying communities and central actors, both important tasks for analysis.…”

Exploring the Possibilities of Embedding Heterogeneous Data Attributes in Familiar Visualizations

“…Hybrid approaches can offer two ways for exploring the same relational data. NodeTrix [17] and MatLink [18] are wellknown hybrid visualizations that take advantage of matrix and node-link diagrams for representing the relations in network data. OntoTrix [3] also combines node-links and matrices to represent ontologies.…”

Hierarchical Visualization of Cloud-Based Social Network Service Using Fuzzy

“…현재 소셜 네트워크 시각화를 위한 연구의 대표 적인 접근방법으로는 노드 링크(NL, node-link) 접 근방법 [3], 행렬 그래프(MAT, matrix graph) 접근 방법 [4], 노드 링크와 행렬 그래프의 혼합형 접근방 법(hybrid of NL and MAT) [5,6]이 주로 연구되고 있으며, 이외에도 다양한 기법들을 기존방법들에 적용하여 성능을 향상시키는 확장형 접근방법들이 있다 [7][8][9] . 비용이 많이 드는 문제를 가지고 있다 [3,5,6] . [3,4] .…”