From Pretend to Expected Contexts

Although simple data types can be used to model application specific context elements that have no explicit representation in the prototype, the software obviously cannot automatically ascertain the current value of this element. A feature that was present in the first prototype, pretend context elements, was intended to provide a substitute for this auto-capturing facility. Pretend context elements allowed the user to manually enter a current value for a context element that was then used as a simulated current value. These pretend values could also be used for context elements that were normally established automatically, e.g. for pretending a current location when no GPS device was attached.

Chapter 3: Ecology Fieldwork Tools

The feature was removed from the second prototype because the effort to adjust the values of different fields through a pretend context dialog every time they changed outweighed the utility that was gained (note that the user effort lies in a combination of the interface design and the nature of the task itself). In many of the example applications considered, very few pretend contexts remained at a constant value; most changed rapidly, e.g. for each giraffe observation completed the location and many other fields would differ. Even for the few values that did stay constant for longer than one stick-e note authoring cycle, having to flip to the pretend context elements’ screen in order to change the values resulted in little or no time saving over simply typing the value directly into each individual stick-e note (though perhaps there are better ways to construct this user interface in this regard). However, experience working with the system in the Kenyan trial led to another concept of how pretend contexts could be implemented that could lessen the burden of their definition and maintenance whilst enhancing the utility that they offer. Rather than simply specifying a single pretend value for a context element, the proposed system is based on the concept of having a selection of methods that can establish an expected value for a context element. Each context element field of a stick-e note is assigned one of these methods for predicting the next value. Some proposed methods for establishing the expected value (encompassing the pretend value and automatic extraction methods currently employed by the prototype) are listed here:

By Static Value. This is the original pretend context method of operation, whereby the user manually enters the current value of a context element. This value is automatically inserted into any appropriate new stick-e note field. By default this will be the selected method for establishing an expected value.

By Adjusted Value. This is an enhanced version of the static value method where, should the user change the value that was automatically inserted into the stick-e note field, then the pretend value is replaced with this newly adjusted value. This eliminates the main problem of taking too much time to adjust pretend context elements. The user would no longer need to flip to another dialog to adjust the pretend value; it is updated as part of the process the user has to carry out anyway.

By Most Used Value. This method works by treating the most commonly entered value as the current pretend context value. For example if I have an ‘Animal Spotted’ field, and the most commonly observed animals have been giraffe, then ‘Giraffe’ will automatically be entered into the next note created.

By Predicted Value. Some context elements are likely to have identifiable trends that can be extrapolated in order to predict what the next current value will be. For example, when walking a transect and taking samples every 100 metres, a detectable sequence of values for location will occur.

By Sequence. Certain fields may be known at the outset to have a particular sequence to them, and these sequences could be specified in advance by the template designer. For example, if recording stick-e notes to tag photographs to a location, the photo number field could be assigned an increment sequence pattern so it is automatically incremented for each new photograph.

By Average Value. For numerical based data, this attempts to estimate the new current context value by calculating the average of those collected so far.

Automatic. This is used for fields where the handheld computer can automatically establish the current context element’s value, e.g. by consulting the GPS receiver to establish the current location.

Blank. In some cases having values that automatically appear can actually waste time instead of saving it, e.g. if the user has to erase the current value before entering the new one. Therefore, if there is no pattern to establishing the expected value automatically the field can be left blank when the stick-e note is created. The emphasis of this new concept is in providing a variety of methods to establish the

expected value rather than simply providing two polarised options based on automatic

extraction from a sensor or manually defining a static pretend value. Some of the suggested new methods exploit a knowledge of the history of the context values that have been recorded in previously created stick-e notes in order to calculate an expected value. Unlike the previous pretend context system, which was appended to the main program, the expected values system should be an integral part of specifying a template.

Chapter 3: Ecology Fieldwork Tools

An expected value method would always be chosen for a field, even if it is the blank option.

The application of the expected values is focused on providing current values for different context elements with as much automation as possible. This is a much narrower focus than the original pretend values that were also used as a mechanism for previewing notes by allowing the user to pretend they were in a certain situation. Such a use led to the need to support multiple values of pretend contexts, e.g. to show stick-e notes for both Canterbury and Nanyuki. Separating the provision of current pretend values from the previewing of notes (new methods for which are discussed later) allows for a simplification (i.e. no multiple values) that makes possible the more powerful concept of expected values.