Proceedings: GI 2013

FacetClouds: exploring tag clouds for multi-dimensional data

Manuela Waldner , Johann Schrammel , Michael Klein , Katrín Kristjánsdóttir , Dominik Unger , Manfred Tscheligi

Proceedings of Graphics Interface 2013: Regina, Saskatchewan, Canada, 29 - 31 May 2013, 17-24

DOI 10.20380/GI2013.03

  • Bibtex

    author = {Waldner, Manuela and Schrammel, Johann and Klein, Michael and Kristj{\'a}nsd{\'o}ttir, Katr{\'i}n and Unger, Dominik and Tscheligi, Manfred},
    title = {FacetClouds: exploring tag clouds for multi-dimensional data},
    booktitle = {Proceedings of Graphics Interface 2013},
    series = {GI 2013},
    year = {2013},
    issn = {0713-5424},
    isbn = {978-1-4822-1680-6},
    location = {Regina, Saskatchewan, Canada},
    pages = {17--24},
    numpages = {8},
    doi = {10.20380/GI2013.03},
    publisher = {Canadian Human-Computer Communications Society},
    address = {Toronto, Ontario, Canada},


Tag clouds are simple yet very widespread representations of how often certain words appear in a collection. In conventional tag clouds, only a single visual text variable is actively controlled: the tags' font size. Previous work has demonstrated that font size is indeed the most influential visual text variable. However, there are other variables, such as text color, font style and tag orientation, that could be manipulated to encode additional data dimensions. FacetClouds manipulate intrinsic visual text variables to encode multiple data dimensions within a single tag cloud. We conducted a series of experiments to detect the most appropriate visual text variables for encoding nominal and ordinal values in a cloud with tags of varying font size. Results show that color is the most expressive variable for both data types, and that a combination of tag rotation and background color range leads to the best overall performance when showing multiple data dimensions in a single tag cloud.