Week 11 · Constructed teaching example inspired by Nobre et al. (2020)
Question: Which European person or institution has the fewest likes?
Click an attribute heading on the right to sort. Click it again to reverse the order. The account’s attributes, matrix row and matrix column move together.
Click an account name to keep it selected while sorting. Hover over a matrix cell to shade its whole row and column.
Bars start at a common baseline within each attribute column. Their scales stay fixed when sorting; the numbers give exact values. Sorting changes the display order; every link and attribute stays attached to the same account.
All account names, counts, categories and links on this page are invented. Each node is a fictional account; each undirected, unweighted link is an invented interaction. There are no self-links. Following counts accounts followed; followers counts accounts following; tweets counts posts; likes is the assigned count used for the question. These four account attributes are independent of this small graph’s degrees.
The paper places an attribute table beside a matrix and keeps the table rows aligned with matrix rows. Its matrix supports sorting by attributes, labels, neighbourhood and a seriation algorithm (an algorithm that finds a row order). This page illustrates sorting by attributes. It deliberately uses the same account order on both matrix axes and provides an added continent-then-likes sort. It is a reconstruction of this interaction with different data, rather than the original study application or an experiment on participants.
Reference: Nobre et al. (2020), Evaluating Multivariate Network Visualization Techniques Using a Validated Design and Crowdsourcing Approach, matrix design and supplementary Task 2. Paper · Authors’ application repository.