The Use and Applicability of Timeline Interfaces
Abstract
A common method of representing data is as a collection of attributes. An entry
in a bibliography might consist of a title, an author, a publication date, and an ISBN
number. A study of the distribution of certain pollutants in a body of water might rely on
a set of water samples, where the information recorded for each includes a location, a
temperature, and the concentrations of various compounds & organisms. Entries in a
relational database conform to this model -- a database is composed of tables, each of
which contains entries which consist of a set of well-defined attributes. Entries in a table
describing individuals might consist of a name, weight, eye color, and mailing address.
This conception of data is very simple, broadly applicable, and can be very flexible -- if
attributes are allowed to reference other data points, it becomes possible to construct
trees, graphs, as well as a wide variety of other complex structures.
This model for data will be referred to henceforth as the attribute-collection
model. Given its simplicity and wide applicability, the value of the attribute-collection
model seems self-evident. But the general nature of the model -- the very quality which
gives it worth -- presents a difficulty when it comes to manipulating data. Particular
structures which might be constructed from data in this form are suited to particular
interfaces. An appropriate interface for manipulating family trees, consisting of
individuals having two parents and zero or more children, might be entirely useless when
applied to a different data-set. The bibliography example mentioned earlier does not
form a tree, and thus cannot be viewed or manipulated in a meaningful way if conceived
of as a tree.