TagRiver is a novel tag-cloud visualization which incorporates both multivariate and temporal data. TagRiver displays tags associated with a number of data elements simultaneously by visualizing the temporal changes within the tag information and the volume of the tags associated with each element. Tags for respective elements are scaled based on their popularity, and then are displayed in vertically adjacent polygonal regions. The area of each of these regions is proportional to the volume of the tag information of the corresponding data element. To be able to distribute various sized tags within a polygonal region we utilize a novel 2D packing algorithm which satisfies certain aesthetic criteria and runs in real-time. TagRiver updates the polygonal area and tag information for each time step and provides a summary visualization for the past time steps, where users only see the volume of the tags associated with each data element. Users are equipped with simple interactive widgets which they can use to traverse and insepct tags associated with any time step on demand.

Tag River was developed in collaboration with Basak Alper. It was presented at the Workshop on Media Arts, Science and Techology 2009 in Santa Barbara, CA.

[git repository]
[MAST 2009]
[Tag River short paper.doc]
[Basak Alper]