Interaction techniques

Visual Analytics for Linguists

ESSLLI 2014

Chris Culy

7 August 2014

Interaction is:

  • the key to success for exploration
  • often important for comprehension/analysis
  • can be useful for communication/presentation

Information seeking mantra

Overview first, zoom, details on demand

General techniques (1)

  • Tooltips
    "title" attribute in html/svg, lots of other tools, e.g. jQuery plugins, other libraries, etc.
  • Pan and zoom
    An issue: how to do pan and zoom.
    • Panning: panning by dragging (but cf. problem with DoubleTreeJS)
    • Zooming: scrollwheel or "flick", but not all computers have these
    • A control could be an alternative means. e.g. in maps: zoom only or both pan and zoom

General techniques (2)

  • Animation
    Animation can be useful in helping users keep oriented when visualization changes. See some discussion and a D3 example here
  • Fisheye
    Mousing over or selecting an area makes it larger and/or other areas smaller. The level of interest determines the scaling. Explanation and D3 examples here

Techniques more specific to InfoVis (1)

  • Semantic zoom (aka level of detail)
    Different amounts or types of information are shown depending on the level of zoom. A map example No roads are shown at this zoom. The more you zoom in, the more roads (and other features) you see.
  • Lens
    A lens provides more detail (or another view) for a selected area of the visualization
    An overlay example (Java applet) and more like that one (Flash). One with multiple lenses, more like the original idea from Xerox PARC

Techniques more specific to InfoVis (2)

  • Coordinated views
    Multiple views of the same data, where changes via interaction in one view are reflected in the other view(s). A graph example (Java). Voyant tools uses this technique.
  • Focus + context
    A type of coordinated view where the user can "zoom" on a part of the data and still keep track of the big picture. Simple D3 example
  • Brushing and linking
    Selecting (brushing) some data selects the same/similar/related data, either in the same view or in coordinated view.
    See pages 3 and 4 here. Another one here

For some more information see the infovis-wiki