. .

Inhalt

Adaptive Directness-
A product oriented approach to evaluate specific HCI Situations

 

Authors: Thomas Kahlisch | Wolfgang Wünschmann

Address: Dresden University of Technology |  Institute for Information Systems | 01062 Dresden

Phone: +49 351 463 8467 | Fax: +49351 463 8491

e-mail: kahlisch@dzb.de | wuenschmann@inf.tu-dresden.de

 

Abstract

1. Introduction

2. Adaptive Directness

3. Usability evaluation of specific HCI Situations

4. Model for generalization of Adaptive Directness

5. Perspective commants

Bibliography


Abstract

Assistive technology for people with special needs is very often combined with a need of additional effort for managing Human Computer Interaction. Different strategies for handling the user interface in such cases have to be compared for finding the best way towards high ergonomic quality. Standardized methods for usability testing have to be adapted to special needs of special users.

For this purpose a new approach is described to define an objective product oriented measure, the Adaptive Directness. This measure has been used to evaluate a prototype of a table browser for blind users. Based on the related results a model has been developed for supporting the generalization or modification of the definition of Adaptive Directness. Hints at open problems and important next steps are given.

 

1. Introduction

There are a lot of reasons for the need of testing the usability of technical products and services. Reffering to assistive technologies the most important reasons are:

  • quality improvement in general

  • comparison of different strategies

  • analysis of week points in training programs

Electronic documents have an increasing importance in all areas of social life (education, occupation, leisure). That is the reason that usability tests for technical aids helping people to get access to electronic documents have an extraordinary importance. In contradiction to the demanding situation there are no in general accepted methods and models for a quantitative description of the influences of assistive technologies on the usabilityof electronic documents for people with special needs. Based on the experience of the authors exist the following reasons for the difficulties of developing evaluation methods for specific HCI situations, using assisitive technology like Braille displays or synthetic speech:

  • I. The number of well trained and experienced blind computer users is small. And assisitive technology is expansive.That makes user oriented field testing methods difficult to use.

  • II. Assistive Technology causes complex dialog situations. An evaluation method, which can handle this complexity, needs to distinguish between the following aspects:

    • problems belonging to the application (product),

    • definition of the task, that has to be solved by the user,

    • influences on the user constitution and

    • the specific problems of using assisitive technology (e. g. limitation of an character based user interface).

This paper describes a formal product oriented approach to evaluate specific HCI situations, using assisitive technology.The idea of this concept bases on the interaction room concept developed by Rauterberg [6]. The primary goal of the authors approach to define an objective measure for evaluating assistive technologies has been focused on contributions to improve the access to electronic documents. From this point of view HCI methods used by sighted persons and those used by blind persons have been analized in detail. The aim to find a concentrated expression to summarize the findings of the comparison has lead to the definition of Adaptive Directness. Extrapolations from practical situations of blind students seem to justify the expectation this measure could be used in very different dialog situations based on assisitive technologies.

2. Adaptive Directness

The current paradigm of assisitve Technology like screen readers gets the blind computer user in a 'controlled by second order' interaction situation. The bridge software handles the information interchange between the user and the computer system. Because of the limitation of the tactile or speech oriented adaptive displays, the bridge software needs to have flexible control mechanisms, which allow the blind user to explore the richness of the available information on the graphic user interface. In most cases the exploration process is necessary and costs plenty of time.

By using Rauterberg 'Interaction Point Model' it is possible, to quantify the number of elements of a given interaction context which are available to the sighted user. By using this method, it is also possible, to quantify the number of elements which are available in the same situation to the blind user.

Rauterberg names the following elements of an interaction context:

  • Interaction points are elements of an user interface like mouse sensitive arrays, menu items or hot keys.

  • Objects are elements which can be changed by interactions (e. g. windows, forms or pictures).

To determine the relationship of the number of elements available to the sighted user to the number of elements available to the blind user the following expression can be calculated for each interaction context:

AR = #IPF / #IPD (1)

AR adaptive relationship of interaction points for a given interaction context

#IPD number of interaction points in a given interaction context
#IPF number of interaction points available to the blind user in a given interaction context

The same relationship can be calculated for the objects in a given interaction context as well:

ARO = #OF / #OD (2)

ARO adaptive relationship of objects in a given interaction context

#OD number of objects in the given interaction context
#OF number of objects available to the blind user

To get full control over a dialog situation the blind user usually needs to explore a given interaction context by using particular reading functions provided by the assisitive technology. The expenditure of this exploration process causes in a product oriented approach in the following items:

  • the sights of the adaptive display and

  • the cleverness of exploration functions.

In an empiric study the following equation was found to calculate this expenditure:

EXP = C**MS (3)
EXP expenditure for a blind computer user to get full control over a given interaction context
C empiric determined factor to rate the sights of the adaptive display and the cleverness of the exploration process
MS minimal number of exploration functions

By combining the Adaptive Relationship of interaction poing equation was found to calculate this expenditure:

EXP = C**MS (3)
EXP expenditure for a blind computer user to get full control overa given interaction context
C empiric determined factor to rate the sights of the adaptive display and the cleverness of the exploration process
MS minimal number of exploration functions

By combining the Adaptive Relationship of interaction points and objects with the EXP function it is possible, to create an product oriented approach to quantify the difference between the information available for a sighted user to a blind user. This measure is called 'Adaptive Directness'.It can be calculated for a dialog system in the following way:

AD = SUM[d=1;n] ARd *EXPd (4)

The Adaptive Directness for the objects of an dialog system can be calculated as follows:

ADO = SUM[d=1;n] AROd * EXPd (5)

 

3. Usability evaluation of specific HCI situations

The relevance of the described approach was check out in a usability evaluation of a particular software system developed at the Dresden University of Technology. The software is a prototype of an special Table Browser for blind computer users. The program provides blind students access to complex tables in hypertext based study materials. The features and functionalety of the browser is described at [5].

By using the described approach, other formal description methods [7] and special KLM [1] operators it is possible, to evaluate all components of usability (effectiveness, efficiency and user satisfaction) [4] for a particular dialog system.

The evaluation process of the Table Browser bases on solving three different example tasks. The steps which have to be made by the user and the belonging results are displayed in tables and other formal descriptions like interaction graphs. By using this formal product and task oriented description method, it was possible, to prove, that the evaluated software system provides blind computer users presentation and navigation functions to read complex tables in different ways.

A more detailed evaluation was supported by using the following special KLM operators:

 

Rb-Operator:
The time to read one Braille character from a Braille display (Rb=0.28)
Hb-Operator:
The time to move the hand from the keyboard to the Braille display and

 

backward (Hb=0.98).

The special operators have been detected in a self test by the author. The results are very close to earlier published values [2].

The results of the KLM have proved the following assumptions:

  1. The time for reading represents the highest expenditure for a blind computer user during an interaction situation.

  2. Well designed presentation and navigation functions are useful to reduce the reading expenditure.

Further research will show whether the described approach can be used for different specific HCI situations and how the product and task oriented model can be expanded with user oriented methods.

4. Model for generalization of Adaptive Directness

4.1 Proposal on a model structure

For further investigations on the usefullness of the measure Adaptive Directness a model as shown in figure 1 should be used.

 

Model for generalization of the concept of "Adaptive Directness"

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1: Model for generalization of the concept of "Adaptive Directness"

 

4.2 Importance of electronic documents

Figure 1 is closely related to the by the authors investigated situation of access to electronic documents. They can in general have very several complexity physical representation. They follow the tendency to become in the near future more and more a substitute of real existing objects. Electronic documents include both components with descriptive content and components with instructive content (description of tasks). Instructions are sometimes given implicitly ore are based on conventions (e.g. reading a book). Consequently the electronic document is combined with an author (generator of descriptive content) and with an instructor (generator of tasks) in figure 1. The parallel situated process interface is an interface to generalprocesses of production or to other life activities substituting other authors and instructors.

Both types of generators influence the level of interaction barriers and in this way the level of Adaptive Directness and in the end the usability of the technical system.

 

4.3 Transformation of interaction flux

Processing of documents needs always tools (combinations of hardware and software). The standard user (e.g. a sighted user) manipulates the document by managing the interaction flux I1, a string of combined input-output activities integrated with planning and decision activities.The user with special needs (user of assistive technologies) has access to the electronic document and to the tool by an adaptor (technical aid) under transformation of the interaction flux I1 into the interaction flux I2. Basis of both interaction fluxes are the interaction points #IPD, #OD, #IPF, #OF integrating additional interaction steps for managing the transformation.The presentation of the adapter in figure 1 with one path to the electronic document and another path to the tool shall emphasize the possibility to integrate very different strategies of assisitive technologies. For instance technologies exist that activate the given document not directlybut by using an on-line or off-line converted version of the document. Therefore additional effort is necessary what has to be taken into account for calculating Adaptive Directness.

4.4 Task and user based approach

The expressions of equations (4) and (5) include all accessible interaction points in an equivalent manner. With the introduction of the task depending interaction flux I1 a task related application of Adaptive Directnes has been prepared. The task reference can be carried out by a task depending vector VT (#IPO, #OD). This vector can be explained as a histrogram function. Very often the time necessary for performing a task is used as a component of usability of the product under consideration. Taking into account a user specific vector of time periods to perform elementary interactions the performance of a complete task can be discussed task and user specific with reference to Adaptive Directness.

4.5 Satisfaction

The functional relation between satisfaction S of an user performing a task and the amount of additional interaction steps IS necessary to perform the same task by using assisitive technologie S = f(IS, AR, AO,..) is very complicated. The satisfaction can be analized in standard situations by using an extensive set of scaled criteria [4]. To decrease the expense for such evaluation methods empiric functions should be introduced comparable with the approach of equation (3). But there is only negligible knowledge up to now on typical structures of such functions for typical applications of assisitve technologies. The cultural context of usage assisitve technologies will influence the empiric functions.

5. Perspective comments

The measure Adaptive Directness can be used to formulate statements on the quality of assisitve technologies. The comparison of accessability of interaction points at the user interface of a tool for handling electronic documents with and without usage of assisitve technologies is the basic principle for the definition of Adaptive Directness.The given prposal of a model for expanding the product oriented approach to a task and user oriented measure can help to expand the knowledge on usability of assistive technologies and can help to harmonize the methods for discussing evaluation results.

The need of comprimation details coming from usabiliirectness.The given prposal of a model for expanding the product oriented approach to a task and user oriented measure can help to expand the knowledge on usability of assistive technologies and can help to harmonize the methods for discussing evaluation results.

The need of comprimation details coming from usability tests to a statement on satisfaction of persons using assisitve technologies will furthermore exist as a serious problem of usability tests. In connection to this problem a very hot topic with high relevance to Adaptive Directness exists in the following question: What are the driving forces for the tendncy that users of assistive technology sometimes change over from doing all necessary interaction steps for having a maximal interaction competence to a management method doing not all necessary interaction steps for full interaction competence and accepting an increased risk of interaction errors.

The challenge to use the measure Adaptive Directness during prototyping of new products belonging to the field of assistive technologies is another very important field for perspective investigations. Puplished guidelines on methods for software development with high ergonomic quality [3] can serve as fundamental knowledge but they have to be adapted to specific needs and rights concerning users of assisitve technologies.

 

Bibliography

 

[1] Card, S. K.; Moran, T. P.; Newell, A.: The Keystroke-Level Model for User Perfomance Time with Interactive
In: Communications of the ACM 23. 1980. S. 327-356.

[2] Denninghaus, E.: Die Foerderung der Lesegeschwindigkeit bei blinden und sehbehinderten Jugendlichen und jungen Erwachsenen.
In: Blind-Sehbehindert, Nr. 2 96, S. 95-100.

[3] W.: International User-Interface Standardization.
In: Allen B. Tucker, Jr. (Ed.): The computer science and engineering handbook.
Florida: CRC Press, 1997, pp. 1474 - 1493.

[4] ISO 9241 Ergonomic requirements for office work with visual display terminals (VDTs)
Part 11: Guidance on usability (Version ISO/DIS 9241-11.2.1997)

[5] Kahlisch, Th.: Making hypertext based study material accessible and user friendly to blind students,
Sixth International WorldWideWeb conference. Accessibility Track. Santa Clara 6. April 1997.

[6] Rauterberg, M.: Ein Konzept zur Quantifizierung software-ergonomischer Richtlinien.
Zürich 1995. ISBN 3-906509-11-7.

[7] Wünschmann, W.: Grundlagen der Mensch-Maschine-Kommunikation, Vorlesungsskript.
TU Dresden. Fakultaet Informatik, Sommersemester 1997.