CS 160: Lecture 5

Professor John Canny

Fall 2001

Sept 11, 2001

Administrivia

• You’re in a project group now (see the web site)

• Make sure you meet this week and come up with a project proposal.

• Hand in the proposal by Friday 5pm.

Task Analysis

• A technique for analyzing existing tasks by observation.

• Doesn’t require understanding ofUsers’ goals, just what they do.

• But because of that its harderto apply to the design of a newsystem.
  • Good for training materials anddocumentation

Task Analysis: 3 Approaches

• Tasks decomposition: looks at how a task is split into subtasks and the order in which these are performed.

• Knowledge-based techniques: what do users need to know about the objects and actions involved in a task? How is that knowledge organized?

• Entity-relation-based analysis: an object-based approach, identify objects, relationships and actions.

Task Decomposition

• Break the task into subtasks:

• Hierarchical Task Analysis (HTA):
  • Organize tasks into a hierarchy
  • Include ordering constraints
  • Looks something like logic programming (PROLOG)

Clean house

Get vacuumcleaner

Cleanrooms

Emptydustbag

Put everythingaway

Cleanhall

Cleanliving room

Cleanbedrooms

Task Decomposition

0. In order to clean house

• Get vacuum cleaner out
• Fix attachment
• Clean the rooms
  3.1 Clean the hall
  
  3.2 Clean the living rooms
  3.3 Clean the bedrooms
• Empty the dustbag
• Put the vacuum cleaner away

Plan 0: Do 1-2-3-5 in that order

Plan 3: Do any of 3.1, 3.2, and 3.3 in any order depending on which rooms need cleaning

Task Decomposition

• Real tasks have more structure:
  • Possible ordering of steps
  • Frequency of steps
  • Task achievement (all or some subtasks)

Stopping Conditions

• P x C rule:
  • P is probability of a user error on the subtask
  • C is the cost of the error

• When P x C is small enough, we stop expanding

• Can also stop at “skill” steps, where the users actions are not reflective (conscious)

Task Variations

• Optional tasks

• Waiting for Events

• Cycles

• Time-sharing

Aside: situated action

• Not all activities fit the task decomposition hierarchy nicely

• Think about tidying a room (vs. cleaning)
  • See an item, put it away
  • Look at the arrangement of things, change it
  • The arrangement of things suggests classifications...

• Your behavior is heavily influenced by the environment

• So don’t forget to observe real behavior when doing task analysis

Knowledge-based analysis

• Basic idea: start with list of items, then organize into a taxonomy (often ends up being a hierarchy)

• e.g. driving:
  • Steering wheel
  • Seat belt
  • Horn control
  • Seats
  • Windscreen
  • Speedometer
  • Dashboard
  • Steering column
  • Foot pedals...

Knowledge-based analysis

• Controls:
  • Steering wheel
  • Horn control
  • Speedometer
  • Foot pedals…

• Structural
  • Seats
    • Seat belt
  • Dashboard
    • Steering column
  • Windscreen

Knowledge-based analysis

• Controls:
  • Steering wheel
  • Horn control
  • Speedometer
  • Foot pedals…

• Structural
  • Seats
    • Seat belt
  • Dashboard
    • Steering column
  • Windscreen

Producing a taxonomy

• Start with an exhaustive list

• Using existing taxonomies (e.g. parts catalog)

• Put the items on cards - have users arrange them

• Taxonomies can be captured with attribute - value pairs, some required, some optional

• TAKD (Task Analysis for Knowledge Description) produces a Task Descriptive Hierarchy (TDH)..
  
  

TDH

• Conjunctives:
  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
  

More on TDH

• TDH can be applied to actions as well as objects.

• In comparison with HTA (Hierarchical Task Analysis), TDH Focuses on similarity of objects

• HTA groups objects which are related by the task

• HTA allows arbitrary repetition, TDH encourages uniqueness.

KRG

• Once we have a TDH, we can describe tasks in quasi-english sentences, using the TDH terms as standard nouns and verbs.

• Idea is to describe the task at the right level of abstraction, with unambiguous terms.

• Sounds a lot like XML...

Entity-relationship techniques

• Borrowed from OO databases

• Objects
  • Concrete Objects
  • Actors
  • Composite Objects

• Actions
  • Have at least a “patient” X (action is done to X)
  • Often an “agent” (who does the action)
  • May involve other objects “instruments”
  • (agent) Sam planted the leeks (patient) with his spade (instrument)

Objects

• Objects
  • Concrete Objects
  • Actors
  • Composite Objects

• Composite objects are collections of objects “the team”, “the kit” - sets of things

• Composite objects may also be “tuples”, e.g. an office comprises desks, chairs, file cabinets...

Actions-messages

• Messages are a very important kind of action because they often influence other actions
  • People ask people to do things
  • People commit to doing things
  • We acknowledge completion of actions

• Speech-act theory studies this

Entities and relationships

Object SamActions: S1: drive tractor S2: dig carrots

Relations: object-object

location(pump3, glasshouse)

Relations: action-eventbefore(plant seed, germinate)triggers(water seed, germinates seed)

Sources of information

• Existing documentation

• Observation in a worked example

• Interviews (create specific scenarios)

Uses of task analysis

• Manuals and documentation

• Requirements capture for design

• Detailed UI design

Summary

• Find your group members!

• Project proposals due Friday (JFC’s office mailbox)

• Task analysis is a way to understand work

• There are 3 different ways of approaching it (which can be used together to some extent)