Tracking Continuously Changing Quantities (NOT PUBLISHED)

In many traditional planning systems, events have been perceived as occurring instantaneously. However, a more realistic characterization of the world is one in which events are considered processes continually changing over periods of time. This paper describes a temporal database system extended to represent continuously changing quantities. In this model, a fact is represented by time tokens corresponding to periods of time in which the fact is believed to be true. A fact may be used to represent a continuously changing quantity by associating with each of its tokens an initial value and a function by which the value changes over time. The time token is said to "track" the value of the quantity that it represents. A transaction on such a quantity (i.e., an event that effects its value) is represented by a time token corresponding to the period of time in which the value is affected and by a function describing the transaction's effect over time.