CMPUT 301: Lecture 15 Task Analysis

1
CMPUT 301 Lecture 15Task Analysis

• Lecturer Martin Jagersand
• Department of Computing Science
• University of Alberta
• Notes based on previous courses by
• Ken Wong, Eleni Stroulia
• Zach Dodds, Martin Jagersand

2
Today

• Some notes on the main course project.
• Hierarchical task analysis
• Temporal execution plans
• Knowledge based analysis

3
Task Analysis

• Process of analyzing the way people perform their
  jobs
• the things they do
• the things they act on
• the things they need to know
• Focuses on observable aspects of a task (as
  opposed to internal cognitive states)

4
Goal

• For Computer Science
• To understand how people solve tasks
• To be able to map parts of this work effectively
  into a computer program (which people still
  usually interact with)
• But nothing new under the sun Consider how early
  capitalist theory (e.g. Adam Smith) and
  industrialists (e.g. Henry Ford) subdivided and
  analyzed factory work to increase economic gain

5
Task Analysis

• To clean the house
• get the vacuum cleaner out
• attach the appropriate tool
• clean the rooms
• when the dust bag gets full, replace it
• put the vacuum cleaner and tools away

6
Task Analysis

• To vacuum, need to know about
• vacuum cleaners
• tool attachments
• dust bags
• storage closet
• rooms to be cleaned
• etc.

7
Task Analysis

• Three approaches
• task decomposition (hierarchical and temporal)
• look at splitting a task into subtasks and the
  order in which these are performed
• knowledge-based analysis
• look at what users need to know about objects and
  actions involved in a task
• entity-relation-based techniques
• look at identifying the actors and objects, the
  relationships between them, and the actions
  performed

8
Task Analysis

• Purpose
• for new systems, helps in capturing and deriving
  the requirements, prior to design
• for existing systems, helps in clarifying
  knowledge about the current situation
• Issues
• may include non-computer activities
• focus on observable user behavior

9
Task Decomposition

• Hierarchical task analysis (HTA)
• produce a hierarchy of tasks and subtasks
• produce temporal plans to describe ordering of
  (sub)tasks and conditions under which they occur

10
HTATask Decomposition

• 0 To clean the house
• 1 get the vacuum cleaner out
• 2 attach the appropriate tool
• 3 clean the rooms
• 3.1 clean the hall
• 3.2 clean the living room
• 3.3 clean the bedrooms
• 4 empty the dust bag
• 5 put the vacuum cleaner and tools away

11
HTATask Decomposition

• Possible temporal plans for tasks and subtasks
• plan 0
• do 1, 2, 3, 5 in that order
• when the dust bag gets full, do 4
• plan 3
• do any of the rooms 3.1, 3.2, or 3.3 in any order
  depending on which rooms need cleaning

12
Task Decomposition

• Temporal refinements
• plan 3
• do 3.1 every day
• do 3.2 once a week
• do 3.3 when visitors are coming

13
Task Decomposition

• When to stop decomposing subtasks?
• Example Computer assisted training system
• P x C lt threshold
• P probability of making a mistake
• C cost of the mistake
• depends on intended purpose and audience

14
Task Decomposition

• Other stopping rules
• When human is too complex to model
• when the task contains complex motor responses
  (human sensory motor)
• when the task involves internal decision making
  (human cognitive)
• Benefit analysis of time, economic return etc.

15
Task DecompositionAdd sequential and temporal
info
16
Task DecompositionRefine and add iteration

• Note loop!

17
Temporal Task Decomposition

• Types of plans
• fixed sequence
• e.g., make pot
• optional tasks
• e.g., empty pot
• waiting for events
• e.g., when kettle boils

• cycles
• e.g., pour tea
• time-sharing
• e.g., boil water and empty pot
• discretionary
• combinations

18
Knowledge-Based Analysis

• Understand the knowledge needed to perform a
  task
• list all objects and actions involved
• build taxonomies to structure these
• use this understanding for training

19
Knowledge-Based Analysis

• Need to know about car control classification
• steering ? steering wheel, indicators
• engine/speed
• direct ? ignition, accelerator, foot brake
• gearing ? clutch, gear stick
• lights
• external ? headlights, hazard lights
• internal ? courtesy light
• parking ? hand brake, door lock
• etc.

20
Knowledge-Based Analysis

• Taxonomy
• different structures might be possible, depending
  on intended purpose and audience
• e.g.,
• driving versus repairing by mechanic

21
Knowledge-Based Analysis

• Task descriptive hierarchy
• XOR
• object in only one category
• AND /
• object in all categories
• OR
• object in at least one category

22
Knowledge-Based Analysis

• car controls XOR
• wash/wipe AND
• function XOR
• wipe ? front wipers, rear wipers
• wash ? front washers, rear washers
• position XOR
• front ? front wipers, front washers
• rear ? rear wipers, rear washers

23
Knowledge-Based Analysis

• kitchen item OR
• preparation ? mixing bowl, plate, chopping
  board
• cooking ? frying pan, casserole, saucepan
• dining ? plate, soup bowl, casserole, glass

24
Knowledge-Based Analysis

• Uniqueness rule
• must be able to distinguish any two specific
  objects
• e.g.,
• cannot distinguish mixing bowl and chopping board
• can distinguish mixing bowl and plate

25
Knowledge-Based Analysis

• kitchen item AND
• / shape XOR
• dished ? mixing bowl, casserole, saucepan,
  soup bowl, glass
• flat ? plate, chopping board, frying pan
• / function OR

26
Knowledge-Based Analysis

• kitchen item AND
• / shape XOR
• / function OR
• preparation ? mixing bowl, plate, chopping
  board
• cooking ? frying pan, casserole, saucepan
• dining XOR
• for food ? plate, soup bowl, casserole
• for drink ? glass

27
Knowledge-Based Analysis

• Unique path
• each object can be represented by a knowledge
  representation grammar (KRG) term
• e.g.,
• a casserole dish
• kitchen item/shape(dished)/functioncooking,dinin
  g(for food)/
• kitchen item whose shape is dished AND its
  function is cooking OR dining for food

28
Knowledge-Based Analysis

• Taxonomy of actions
• can also structure actions, based on genericity
  instead of a how to do it decomposition
• make general statements about tasks

29
Knowledge-Based Analysis

• kitchen job OR
• preparation ? beating, mixing, pouring
• cooking ? frying, boiling, baking
• dining ? pouring, eating, drinking

30
Knowledge-Based Analysis

• Taxonomies of objects and actions
• crosscheck to look for omissions
• perhaps easier to see what people use than to see
  what they do
• add taxonomies
• e.g., classify foods

31
Knowledge-Based Analysis

• Generic descriptions of simple tasks
• KRG sentence
• e.g.,
• eating a carrot off a plate
• kitchen job(dining)using a kitchen
  item/shape(flat)/functiondining,preparation/to
  food(root vegetable)

32
Knowledge-Based Analysis

• Issues
• generification
• not bothering with certain low-level distinctions
• too many different KRG sentences
• perhaps more generification needed
• too few different KRG sentences
• level of abstraction too great

33
End

• What did I learn today?
• What questions do I still have?