Great to see these issues coming up here! In my PhD research on visual timelines carried out at the Royal College of Art with Stephen Boyd Davis in collaboration with System Simulation we have looked specifically at uncertainties in time, both in the data as well as in the visual representation.
Often data visualisations lack any indication of uncertainty simply because they can't take advantage of date descriptions in natural language, which may help to clarify the scope of recorded dates.
There are some examples of digital timelines which do take uncertainty into account. Most notably Neatline (http://neatline.org/) which expresses uncertainties in time as gradients and SIMILE Timeline, which allows adding a 'latest start' and 'earliest end' property to durational events (but not to non-durational events). The 'uncertain' times are then rendered as translucent blocks (see http://simile-widgets.org/wiki/Timeline_Event_Display).
In the context of medical timelines, where ambiguities and uncertainties are a big issue, there are some approaches to modelling and visualising uncertain times. Some of them we have outlined in a recent paper on the subject: http://ewic.bcs.org/content/ConWebDoc/50996.
An overview of historic visual models of time, which present a greater awareness of uncertainties than current examples, can be found in our workshop paper for a recent CHI workshop on time: http://research.kraeutli.com/wp-content/uploads/2013/11/Boyd-Davis-2013-Time-in-Perspectivea-visual-approach-to-models-of-time.pdf.
There is unfortunately no standard way of visualising or recording uncertainties in time. The problem is often not so much that the timeframe is unclear (i.e. how large a buffer is intended by 'ca.'), but that the context of the estimate doesn't get recorded. A painting, for example, may be dated based on the lifetime of its painter, the person depicted or an external event. Putting that kind of information in plain text in a 'display date' field doesn't really help for (automatically) visualising data. Probably, instead of expressing uncertainties somehow in numbers, it would be more useful to express the reasoning behind dates – something which is possible with Linked Data, but hard to accomplish in traditional database formats.
****************************************************************
website: http://museumscomputergroup.org.uk/
Twitter: http://www.twitter.com/ukmcg
Facebook: http://www.facebook.com/museumscomputergroup
[un]subscribe: http://museumscomputergroup.org.uk/email-list/
****************************************************************
|