MODELING THE TASK

Leveraging Knowledge-in-the-head at Design Time

George Christou

Department of Computer Science, Cyprus College, Nicosia, Cyprus

Robert J. K. Jacob

Department of ComputerScience, Tufts University, Medford, MA, USA

Pericles Leng Cheng

Department of ComputerScience, Cyprus College, Nicosia, Cyprus

Keywords: Human Computer Interaction, Design modeling, Mental model, Interaction Design, Evaluation framework.

Abstract: A key problem in Human Computer Interaction is the evaluation and comparison of tasks that are designed

in different interaction styles. A closely related problem is how to create a model of the task that allows this

comparison. This paper tries to tackle these two questions. It initially presents a structure (Specific User

Knowledge Representation) that allows the creation of task models which allow direct comparisons between

different interaction styles. The model allows the researcher or the designer to evaluate an interaction design

very early in the design process.

1 INTRODUCTION

The revolution that is witnessed in interface design

today, brings an impressive and diverse set of

interaction styles, like Tangible User Interfaces

(TUI) (Ishii & Ullmer, 1997), and many others. All

these new interaction styles are becoming more

varied and much less unified than previous

generations, seemingly without cohesion on which

to allow any modeling These Reality Based

Interaction (RBI) (Jacob, 2004) styles are trying to

mimic real-world manipulations, and draw from the

skills that users already possess in the real world to

allow the user to interact with the computer.

Because of this disparity, it is very difficult to

characterize them and understand their underlying

principles, like it was done for Direct Manipulation

(Hutchins, Hollan & Norman, 1986). We believe

though, that under this seeming disparity, there are

many similarities both in the theoretical bases and

the design approaches of RBIs.

This paper presents a descriptive structure for the

knowledge that RBIs leverage, and for their

specification. The theoretical structure is called

Specific User Knowledge Representation (SUKR)

and it allows the modeling of user “knowledge-in-

the-head” and interface “knowledge-in-the-world.

2 BACKGROUND

2.1 Theoretical Basis

The work presented in this paper is based on many

different theoretical approaches. The basis of the

research for the creation of the model though, is

Task Analysis (TA), and more specifically,

Cognitive Task Analysis (CTA). Diaper (2004)

defines TA as “the collective noun used in the field

of ergonomics, which includes HCI, for all the

methods of collecting, classifying and interpreting

131

Christou G., J. K. Jacob R. and Leng Cheng P. (2006).

MODELING THE TASK - Leveraging Knowledge-in-the-head at Design Time.

In Proceedings of the Eighth International Conference on Enterprise Information Systems - HCI, pages 131-134

DOI: 10.5220/0002492301310134

 SciTePress

(a) (b)

Figure 1: Task Diagram of creating a blue-filled rectangle in MS Paint™ (a) and TUIDraw (b).

data on the performance of systems that include at

least one person as a system component” (Diaper,

2004, p.14). CTA is defined by Chipman, Schraagen

and Shalin (2000) as “the extension of traditional

task analysis techniques to yield information about

the knowledge, thought processes and goal

structures that underlie observable task

performance” (Chipman, Schraagen & Shalin, 2000,

p. 3). CTA theories provide specific methodologies

for gathering and analyzing the appropriate data.

They begin with a study of the jobs involved in

order to determine which tasks should be analyzed

(Chipman et al., 2000). The second step in CTA is to

identify the knowledge representations that need to

be used (Chipman et al., 2000). The final step is to

use knowledge elicitation techniques that apply,

based on the CTA theory of choice, since there exist

many (Diaper & Stanton, 2004)

2.2 Task Knowledge Structures

Task Knowledge Structures (TKS) (Johnson &

Johnson, 1991) is a theoretical and methodological

approach to modeling tasks. It is a method of CTA

that assumes that when people learn declarative and

procedural facts that pertain to the same topic, the

knowledge is not stored as stand-alone facts. Rather

knowledge is grouped in coherent wholes, so that it

can be recalled and used as a unit. TKS includes not

only knowledge about actions, but also about objects

used to perform those actions. In this way it falls

under the external cognition theory proposed

Norman (1988) and later by Scaife and Rogers

(1996) which is discussed in section 3. TKS were

designed to be a tool for design generation. By

modeling user knowledge a designer can use the

theory to generate design solutions for interactive

systems (Hamilton, Johnson and Johnson, 1998).

The presented model is based on the same

assumption of Johnson and Johnson (1991)

Hamilton, Johnson and Johnson (1998) talk

about objects in TKS and hint at the affordances

(Norman, 1988) of objects, the object roles are not

explicitly defined in terms of a user interface, nor

are the affordances and constraints of these objects

included in TKS. The proposed structure extends

TKS with the addition of this information as shown

in section 3.

2.3 Terms

The model uses terminology that was first presented

by Christou and Jacob (2005). The terminology was

created to allow researchers and designers to refer to

ICEIS 2006 - HUMAN-COMPUTER INTERACTION

132

various parts of an interface without resorting to

interaction style specific terms. The three parts that

were identified in a user interface are the following:

1. The artifacts that represent the data that

can be manipulated. We call these the Data

Objects (DOs). The DOs are not the actual data

in the system. They are the interface’s

representation of data in groups that are

understandable and identifiable by a user.

2. The artifacts which are perceived by the

user to be the means of interacting with the DOs.

We call these Interaction Objects (IOs). The IOs

are many times, the means by which the

interaction occurs between the user and the

interface.

3. The actual artifacts that are manipulated

by the user in order to manipulate the IOs. We

call these the Intermediary Objects (INs). The

intermediary objects are, most of the time,

coupled with an IO, and this relationship is

usually constant.

2.4 Bindings

Bindings are the places where the IO and the DO

become connected, in order to carry out an action by

the user. When an IO touches or in some other way

comes connects with the DO, we say that the IO is

bound to the DO.

Bindings reveal the places where interaction is

possible between the user and the DOs. These places

show where interaction is possible between the tools

and the data of the interface. Since these are the

major places where actions take place, it would be

sensible to single them out for study.

Two types of bindings are identified. Static

bindings, where the relationship between two objects

is always there and never changes, and dynamic

bindings, where the binding exists only for a limited

time, usually until the user finishes an action or a

task.

For example, the binding between the mouse and

pointer is static, because it never changes, no matter

what the application does. The binding between an

icon and the pointer during a drag-and-drop action

however, is dynamic, because it only exists during

the time of the action.

3 SPECIFIC USER KNOWLEDGE

REPRESENTATIONS

The model that results from the analysis of the

interface is the Specific User Knowledge

Representation (SUKR). SUKRs are a form of CTA,

and are based on the theory of external cognition

(Scaife & Rogers, 1996). Scaife & Rogers (1996)

postulate that humans do not only use mental models

and mental representations to interact with the

world. They argue that artifacts in the real world are

very much part of reflective behavior. When humans

perform any cognitive task, they use artifacts that

become part of the problem representation, and the

correct user model of the artifact allows for better

solutions to the pertinent tasks.

The model presented here is a task model and not

a user model, thus it does not require any goal driven

behavior. The model describes the task based on

some interaction style, using the interaction style’s

actions, rather than the user’s or researcher’s model

of how the task should be performed. The model

tries to capture the knowledge needed for each task

that can be performed in the interface under some

interaction style.

SUKRs are comprised of two parts, each with a

specific goal:

Pre-Conditions

• Marker use.

Task Performance

1. Place marker on color

a. Marker-to-color Dynamic Binding

2. Place marker on Filled Shape

a. Marker-to-Filled Shape Dynamic

Binding.

3. Place sha

e on drawin

area.

Figure 3: Rectangle Drawing Task in Tangible Use

Interface.

Pre-Conditions

1. Mouse-to-Pointer Static Binding

2. Single Click Action

3. Drag-and-Drop Action

Task Performance

1. Single-click on Rectangle Button

a. Rectangle-to-Mouse

Dynamic Binding

b. Drag-and-Drop in drawing

area to draw shape

2. Single-click on Fill Tool

a. Fill Tool-to-Mouse Dynamic

Binding

3. Single-click on Fill Color

a. Fill Tool-to-Color Dynamic

Binding

4. Single-click inside Rectangle

Figure 2: The Rectangle-Drawing Task in WIMP.

MODELING THE TASK - Leveraging Knowledge-in-the-head at Design Time

133

1. The pre-conditions section, which aims to

capture the minimum amount of procedural

and declarative (but not domain) knowledge

needed by the user to perform the task in

the given interface. The model supposes

that domain knowledge is constant over all

interaction styles, and

2. The task performance section, which

describes the way the user should perform

the action in terms of the DOs, IOs, and

INs, and by including the necessary

Bindings.

Domain knowledge is considered constant

throughout interaction styles and that is why it is not

considered in the SUKR.

The task used to show the modeling procedure is

drawing a filled rectangle in MS Paint™ and in

TUIDraw, a Tangible User Interface Drawing

program, bult by the authors. Fig. 1 shows an

example of a task in two interaction styles. From the

task diagram, the SUKR may be created in the

following way. The diagram is created by breaking

the task up in its constituent actions and each action

is represented by a circle. Any actions that may be

performed in any order are signified by placing their

circles in the same level of the diagram. For each

action the knowledge needed to perform it is

delineated.

The common knowledge for all tasks, such as the

static binding of the mouse to the pointer and that

left-clicking on buttons changes the function of the

pointer is put in the preconditions section of the

SUKR, and knowledge specific to the execution of

the action, and dynamic bindings that occur during

the execution of the actions of the task are put in the

task performance section. The full SUKRs can be

seen in figs. 2 and 3.

4 CONCLUSIONS AND FUTURE

WORK

In this paper the concept of Specific User

Knowledge Representations was presented, along

with the relevant specification method.

Future work that needs to be done is to clarify

and specify the procedure for knowledge elicitation,

and more experiments need to be performed, mainly

with experienced users, to show that the measure

holds not only for novice performance, but also for

intermediate and expert users.

REFERENCES

Chipman, S. F., Schraagen, J. M. & Shalin, V. L. 2000.

Introduction to Cognitive Task Analysis. In Chipman,

S. F., Schraagen, J. M. & Shalin, V. L. (Eds.) 2000.

Cognitive Task Analysis. Mahwah, NJ: Lawrence

Erlbaum Associates

Christou, G. and Jacob, R.J.K. 2003. Evaluating and

Comparing Interaction Styles, Proc. of DSV-IS 2003

Eurographics Workshop on Design, Specification

and Verification Interactive Systems. Springer-

Verlag.

Christou, G. & Jacob, R. J. K. 2005. Identifiability: A

Predictive Quantitative Measure for the Comparison of

a Task Designed in Different Interaction Styles.

Proceedings of HCI International 2005. Lawrence

Erlbaum Associates. CD-ROM.

Diaper, D. 2004. Understanding Task Analysis for Human

Computer Interaction. In Diaper, D. & Stanton, N.

(Eds.) 2004. The Handbook of Task Analysis for

Human-Computer Interaction. Mahwah, NJ: Lawrence

Erlbaum Associates, pp. 5-47.

Diaper, D. & Stanton, N. 2004. The Handbook of Task

Analysis for Human-Computer Interaction. Mahwah,

NJ: Lawrence Erlbaum Associates.

Hamilton, F., Johnson, P., and Johnson, H. (1998) Task-

Related Principles for User Interface Design. Van der

Veer, G.C. (Ed). Proceedings of the Shaerding

Workshop on Task Analysis, Austria

Hutchins, E., Hollan, J., & Norman, D. 1986. Direct

Manipulation Interfaces. In D. A. Norman & S. W.

Draper (Eds.) User Centered System Design: New

Perspectives in Human-Computer Interaction.

Lawrence Erlbaum Associates: Hillsdale, NJ.

Ishii, H. and Ullmer, B. 1997. Tangible Bits: Towards

Seamless Interfaces between People, Bits and Atoms.

Proc. ACM CHI’97 Conference on Human Factors in

Computing Systems. pp. 234-241.

Jacob, R. J. K. 2004. Reality-Based Interaction: A New

Framework for Understanding the Next Generation of

Human-Computer Interfaces. Retrieved October 18,

2005, Tufts University, Dept.of Computer Science

Web Site:http://www.eecs.tufts.edu/~jacob/ theory/

Johnson, H., and Johnson, P. 1991. Task Knowledge

Structures: Psychological Basis and Integration into

System Design. Acta Psychologica, 78 pp. 3-26.

Norman, D. 1988. The Psychology of Everyday Things.

New York: NY. Basic Books.

Scaife, M. & Rogers, Y. 1996. External Cognition: How

do Graphical Representations Work? International

Journal of Human Computer Studies, 45, 185 – 213.

ICEIS 2006 - HUMAN-COMPUTER INTERACTION

134