Task Analysis in HCI

5/31/2020 DeKUT 1
 Content
• Human-Computer Interaction
– What it is; What are its concerns;
– What it is not
– Why the need for HCI and interface
design process
– The process

• Why the need for cross-fertilization

between and coupling of Human
Computer Interaction and Language
Processing
5/31/2020 DeKUT 2
 What is HCI?
A discipline concerned with the
• design,
• implementation and
• evaluation
of interactive computing systems for
human use, their effects on humans and
the study of related phenomena (ACM
SIGCHI)
Hi Mary,
Traffic will be lighter than
normal on your way into
design town.

evaluation implementation

5/31/2020 DeKUT 3
 A major concern…

As a designer it is your important duty

to

Ensure usefulness and usability of

Device or application

5/31/2020 DeKUT 4
 HCI: Concerned with Interfaces and
Interactions
• Interface: the aspect of the system/object
through which the user interacts

• Interaction: any form of communication

between user and system/object
– Typing
– Speaking
– Gestures
– Auditory voice output (sound, voice,
etc.)
–…
5/31/2020 DeKUT 5
 A multi-disciplinary field
• Cognitive psychology and Cognitive
science more generally
• Social psychology
• Ethnography
• Design
• Computer Science
• Artificial Intelligence
•…
5/31/2020 DeKUT 6
 Some of the Questions
Addressed
• What is appropriate for the user’s
environment and the user’s tasks?
• What is appropriate given users’
cognitive abilities?
• What functionalities should be
included?
• How should they be presented?
Accessed? Understood?
• Does it work?
5/31/2020 DeKUT 7
 What HCI is NOT
• Cosmetics
• A last minute “beauty mask”

5/31/2020 DeKUT 8
Why HCI? Why an interface design process?
• 63% of large software projects go over cost
– managers gave four usability-related reasons
• users requested changes
• overlooked tasks
• users did not understand their own
requirements
• insufficient user-developer communication
and understanding

• Usability engineering is software engineering

– pay a little now, or pay a lot later!
– far too easy to jump into detailed design that
is:
• founded on incorrect requirements
• has inappropriate dialogue flow (Saul Greenberg)
• is not easily used
5/31/2020 DeKUT 9
• is never tested until it is too late
Why HCI?
• Machines are increasingly:
– Everywhere: ubiquitous, pervasive, smart
spaces
– Important for every day tasks – nor only for
work anymore
• It is crucial to facilitate and optimise
human/machine interactions, ensuring that the
interaction (and interface) is appropriate for
users’ needs and abilities. This is needed for:
– The success of products
– Their safe and effective use
– Users’ satisfaction (a good user experience)

5/31/2020 DeKUT 10
Efficiency
• Systems often under-utilised (Eason,
1985), in terms of:
– The number of times they are used
– The available functionality
• Error rate, task effectiveness in
terms of time

Time spent at design and errors fixed

at design can avoid very costly
changes later onDeKUT
5/31/2020
in the process 11
Some foundations for designing
interfaces
• Understand users and their tasks
– Task-centred design:
• Task-centred process
• Developing task examples
• Task scenarios and walkthroughs
• Design with the users
– User-centred design and prototyping
• User-centred system design
• Low-fidelity prototyping methods
– Evaluate with users
• Observe people using systems
• Detect inappropriate design and correct, leading to
iterative design

5/31/2020 DeKUT 12
Some foundations for designing
interfaces
• System’s designers/engineers are
not users
• Many “human errors” are actually
errors in design
• People vary in thought and behavior
just as they do physically

5/31/2020 DeKUT 13
Umm, thanks for the warning,
What happens when you
but what should I do?
cancel a cancelled operation?

Do I have any choice in this? Uhhh… I give up on this one

5/31/2020 DeKUT 14
 More Inane Dialog Boxes
•These are too good not to show

5/31/2020 DeKUT 15
 And again…

ClearCase, source-code control Rational Software

5/31/2020 DeKUT 16
• HCI specialists can help by providing
a good conceptual model and filling
gaps between models
• Good design is difficult
• Need to design to accommodate
individual differences

5/31/2020 DeKUT 17
Role of HCI Specialist: Building Bridges
amongst Representations/Models

Mental Mental
representation representation

Designer User

System

Documentation

“Image” of the system

5/31/2020 DeKUT 18
 Some usability principles
(J. Nielsen,

• Relevance of the dialogue: say only what

is necessary, in a logical and simple
order Speak the language of the user –
not the language of the designer/IT
• engineer
• Give the appropriate information when it
is necessary
• Be coherent: in the terminology, in the
types of actions, in the presentation
• Provide immediate feedback
5/31/2020 DeKUT 19
 Some usability principles
(con’d)
• Minimal actions: include short-cuts
• Ensure error messages are clear and
useful; Provide and suggest a way to
recover
• Prevent errors when possible:
anticipate errors through user
testing; avoid “fatal” errors
• Instructions/etc must be clear,
concise and task-oriented.
5/31/2020 DeKUT 20
 Designing an interface
• Iterative Process
• User-Centred, with user participation

5/31/2020 DeKUT 21
 Interface Design and Usability Engineering
Articulate: Brainstorm Refined Completed
• who users are designs designs designs
Goals: • their key tasks

Task
Psychology of Participatory Graphical
centered
everyday interaction screen
system
Evaluate things design Usability Field
design
tasks User Interface testing testing
Methods: Participatory Task scenario
involvement guidelines
design walk-
Representation through Style Heuristic
User-
& metaphors guides evaluation
centered
design

low fidelity high fidelity

prototyping prototyping
methods methods

Products: User and Throw-away Testable Alpha/beta

task paper prototypes systems or
descriptions prototypes complete
specification
5/31/2020 DeKUT 22
 Interface Design and Usability Engineering
Articulate: Brainstorm Refined Completed
• who users are designs designs designs
Goals: • their key tasks

Task
Psychology of Graphical
centered Participatory
everyday screen
system interaction
Evaluate things design Usability Field
design
tasks User Interface testing testing
Methods: Participatory
involvement Task guidelines
design
scenario Style
User- Representation Heuristic
walk- guides
centered & metaphors evaluation
through
design

low fidelity high fidelity

prototyping prototyping
methods methods

Products: User and Throw-away Testable Alpha/beta

task paper prototypes systems or
descriptions prototypes complete
specification
5/31/2020 DeKUT 23
 Task, Context and Users

5/31/2020 DeKUT 24
 Task
• Definition: Goal/Intention of the user and how to
achieve the goal
• Purpose: User-Oriented, not system oriented;
– Not what functionality should the system
have? But
– Who will use the system to do what? and
– How users perform their tasks
• Obtained through task-analysis [there are a
variety of different methods and notations]

5/31/2020 DeKUT 25
 Context
• Physical context: noise, light, stress,
mobility, objects used, safety of
environment, time-criticality, etc.
• Social context: role of information,
communication need, work vs
leisure, etc.

5/31/2020 DeKUT 26
 Task Centered System
Design
Phases:
1. Identification: identify specific users and
articulate their concrete tasks
2. Requirements: decide which of these
tasks and users the design will support
3. Design: base design representation &
dialog sequences on these tasks
4. Walkthrough Evaluations: using your
design, walk through these tasks to test
the interface
5/31/2020 DeKUT 27
 Users
A good design always takes into consideration
human abilities and limitations (physical, mental,
cognitive, etc.)
• Know your users
– Their abilities and limitations
– The characteristics of the targeted user
population
– Their needs
– Their expertise
– The different types that exist in the targeted
user population
5/31/2020 DeKUT 28
 Good Representations (e.g.,
task models, user models)
• captures essential elements of the
event / world
• deliberately leaves out / mutes the
irrelevant
• appropriate for the person and their
interpretation
• appropriate for the task, enhancing
judgment ability
5/31/2020 DeKUT 29
 Representations
•Representations: both for the designers to understand the
problems, and in the interface for the users to support them
in their tasks
• Good representations
– allow people to find relevant information
• information may be present but hard to find
– allow people to compute desired conclusions
• computations may be difficult or “for free” depending
on representations

– Having a set of representations allows one to see

different aspects of the problem
– Solving a problem simply means representing it so as to
make the solution transparent

5/31/2020 DeKUT 30
 Evaluating with users
• Iterative design
– does system behavior match the user’s task
requirements?
– are there specific problems with the design?
– what solutions work?
• Debug designs by observing how people use them
– quickly exposes successes and problems
– specific methods reveal what a person is thinking
– but naturalistic vs laboratory evaluations is a tradeoff
• precision and direct control over experimental design
versus
• desire for maximum generalisability in real life
situations
– Usability engineering approach: observe people using
system in simulated settings (usability labs)
– Continuous evaluations DeKUT
5/31/2020 31
 Cross-Fertilization between HCI and
Language Technology – Why?

• Both concerned with communication as core

concept
• Both attempt to maximise naturalness of
communication for end-user
• Central challenge for both disciplines (at least
for Natural Language Generation within LT):
– Choice and adaptation of appropriate form of
communication for specific user and context
• Increasingly, communication is via language
• Yet, little collaboration between the 2 disciplines
(but increasing…)

5/31/2020 DeKUT 32
 Cross-fertilisation on common
application areas
• Interactive documents (e.g., hypertext and web
pages, hypermedia)
-- “documents as interfaces”
Challenges:
– Providing the right information at the right
time
– Ensuring coherency of hypertext and
appropriateness of the structure
– Avoiding cognitive overload
– Avoiding user disorientation

5/31/2020 DeKUT 33
 Cross-fertilisation on common
application areas (2)
• Natural Language Interfaces: getting
the interaction right
– NL interfaces to databases
– Instructions
– Dialogues systems
– Systems with speech input & output
• Multimodal/multimedia interfaces

5/31/2020 DeKUT 34
• Issues of textual coherence, information
summarisation, selection of appropriate type
language, etc must be taken into account in
interaction design
• Experimental work to validate and characterise
claims concerning efficiency and learnability of NL
interfaces
• Identifying the scope of technology
• Allowing users to understand limitations/scope of
the system.

5/31/2020 DeKUT 35
Cross-fertilisation of
methodological frameworks
• Behavioral and observational
methods
• Task-focused analysis

vs

• Corpus based analysis (manual or

through learning)

5/31/2020 DeKUT 36
 Coupling HCI and LT –
Example
the design and development of an authoring tool to support
technical communicators
– Task analysis done
• Allowed developers to understand the role of the
technology within the documentation process
– Task models used as the input to the text generator
• Input to the generation system is re-usable
• Its construction is spread over its many uses (design,
evaluation, etc.)
– NLP exploited to aid in task modelling/task acquisition
– Usability analysis performed to ensure usability of the
tool by intended target users
– Task-based evaluation of the text produced by system
done to ensure usefulness of output

5/31/2020 DeKUT 37
 …
Isolde – an example. From a task model

5/31/2020 DeKUT 38
 Isolde – … to the instructional text

5/31/2020 DeKUT 39
 Coupling HCI and LT –
Example (2)
• Task-Driven Information Presentation Select and
integrate relevant information for the task
– Provide appropriate information at the right
time
– Highlight most relevant information to allow
operators to focus on important details.

Create a display by reasoning about the role of

the information and its importance – taking a
discourse approach to display design

5/31/2020 DeKUT 40
• Task analysis performed:
– task description and decomposition
– information requirement for each task
– information organisation/relation (to assist the
VDP in presenting information appropriately)
– information source/activity, e.g., information is
found on the on-screen map display
• Mapping of tasks into information needs
• Definition of the discourse structure required to
satisfy these needs

5/31/2020 DeKUT 41
 Selecting and Integrating
Relevant Information into
One Display

5/31/2020 DeKUT 42
 Providing Right
Information at the Right
Time
Display tailored to operators’
tasks and situation leads to
better decision making

All information at any point

in time leads to visual
clutter
5/31/2020 DeKUT 43
 Interface Design and Usability Engineering
Articulate: Brainstorm Refined Completed
• who users are designs designs designs
Goals: • their key tasks

Task
Psychology of Participatory Graphical
centered
everyday interaction screen
system
Evaluate things design Usability Field
design
tasks User Interface testing testing
Methods: Participatory Task scenario
involvement guidelines
design walk-
Representation through Style Heuristic
User-
& metaphors guides evaluation
centered
design

low fidelity high fidelity

prototyping prototyping
methods methods

Products: User and Throw-away Testable Alpha/beta

task paper prototypes systems or
descriptions prototypes complete
specification
5/31/2020 DeKUT 44