Date post: | 21-Dec-2015 |
Category: |
Documents |
View: | 216 times |
Download: | 3 times |
Day 2
Prior to today
• Participated in the one day class on Cognitive Ergonomics, read all material of that class
• studied the reader for today
• read the paper by vdVeer & Mariani
After today
• management plan and time schedule
• interviews with potential users
• video records for interaction analysis
• review relevant information from web and other document sources
• version 1 of Task Model 1
• collect ideas for task model 2
prepare presentations for next time:
• management plan
• task analysis methods and techniques used: why, how, issues
• task model 1
5. Task analysis of complex systems
Overview
5.1. Task analysis in the process of designing complex systems
5.2. Describing current work situations:
5.2.1. Analytic methods
5.2.2. Understanding implicit expert knowledge
5.2.3. Explicit group knowledge (video illustration- Boeing 747-400 accident)
5.2.4. Understanding implicit group knowledge (video ill. Examination system)
5.3. GTA
5.4. Envisioning the future system
5.5. Evaluating task model 2
5.1. Task analysis in the process of designing complex systems
Design of complex systems for users:
the virtual system and the mental model
Design for the mental model:
3 types of design activities• analysis• specification• evaluation
“Task analysis”, in modern approaches, indicates analysis of the whole task domain: business, work, users, stakeholders, situation
“Specification” focuses on details of technology
work organization/practice
Client
users’knowledge/behavior/needs
Technology
Task Model 1
Task Model 2
Scenario
Simulation
Prototype
Functionality
Dialog
Representation
Implementation
usabilitymeasuring
ethnography
psychologicalknowledgeacquisition/hermeneutics
problemanalysis/specification
specification/negotiation
constraints/opportunitiesfeedback
specification
early evaluation
earlyevaluation
UVM maintainingconsistency
Documents/artifacts
validity analysis
As soon as the system is implemented
Task analysis
– describing the current situation
• task model 1
– envisioning the future
• task model 2
Task analysis in literature
Several meanings• collecting knowledge about, and describing, current task domain (task
model 1)• specifying new task situation, envisioning use of novel technology (task
model 2)• specifying details of technology (existing or new) (the user virtual
machine)
approaches always consider type of task knowledge, and relate this to model intended for design process
Literature and approaches often consider only one or two, and often mix up meaning of what is being modeled
Important concepts and distinctions
task and goal:
• each task (activity) has a goal (state to reach)
• a goal may be reached through several different tasks
primary and secondary tasks
• a primary task is generic for the task domain irrespective of technology and situation
• a secondary task is related to the use of tools and technology
basic and unit tasks
• a users’ unit task is an atomic primary task
• a basic task is the simplest task available for delegation to technology
actions
• an action has no goal, its meaning is derived from the task it is part of
5.2. Describing current work situations: task model 1Brigitte Jordan (1996): sources of task knowledge
Explicit expert knowledge
• facts of task domain
• heuristics
• skills
Explicit group knowledge
• process models, rules
• instructions
• historic documents
• anecdotes
Implicit expert knowledge
• “expertise”
• intuition
• common sense
Implicit group knowledge
• “culture”
• COP (community of practise)
• work practice
5.2.1. Analytic methods:
conceptual framework• set of basic concepts (task, goal, condition, action)• relations between concepts
formal representations, based on conceptual framework• task hierarchies (HTA, based on Taylor)• task-action grammars (TAG, ETAG), • goals and operation rules (GOMS, CCT)
planned process• steps in collecting knowledge, modeling, specification
focus on the individual human and his knowledge• cognitive psychology
TKS (task knowledge structure) P. Johnson & H. Johnson (QMW college, London)
conceptual framework
• tasks, plans (task structures), strategies
• procedures (object - action relations)
• roles (sets of tasks assigned to certain individuals)
Psychological basis is conceptual knowledge in long term memory
• psychological techniques: interview, observation, card sorting
• document analysis
TKS (continued)
method:
• from individual TKSs to generic task model GTM (task model 1)
• from GTM to Specific task model STM (task model 2)
• from STM to spec. interface model SIM (functionality at conceptual level)
problems:
• experts often do not know details of others’ roles
• often there is no overview of all roles in an organisation
• objects (things) are only analysed at secondary level
ATOM - Paul Walsh (STC Technology, Harlow, UK)
Start of analysis is identification of “principal objects”
• active objects (agents)
• passive objects (things)
• environment (includes things, agents, technology
objects structured in class hierarchies and relations between types
actions establish a causal link between object states
actions are identified secondary, identified by type of verb:
• process verbs (d.g. “dies”)
• state verb (“waits”)
• action verb (“transports to”)
MAD - D.Scapin, S. Sebillotte (INRIA, France)
Conceptual framework
• tasks, goals of people - states to be reached
• functions of systems (delegated subtasks)
• task hierarchy, subtasks, elementary tasks
• constructors (PAR, ALT, PAR, LOOP, COND, OPT)
• initial state, final state (describe change due to effect of task)
• prior condition, post condition
Formalism
• task / goal tree (HTA)
MAD (continued)
The Graesser method:
• describe as complete and detailed as possible how you …. (name the global label of the task to be described)
“... because ...”
“... Therefore ...”
• keep track of hierarchical structure (need a second analyst)
• filling the gaps
“How ?”
“Why ?”
MAD (continued)
Generic levels of task knowledge
• abstract general task formulation
• specific context sub-tasks (expert task language)
• domain independent procedures: (“store a copy”, “send a message”)
• elementary actions (“ask for signature”, “put date on record”)
5.2.2. Understanding implicit expert knowledge
Implicit knowledge, behavior, skills, “expertise”
• individual knowledge that can not be verbalized
Technique:
• observation, registration
Hermeneutic interpretation
• “what would be my knowledge and mental process, if I would show the behavior that this expert is showing”
Reliability problem
• needs training and multiple interpreters
5.2.3. Explicit group knowledge
Explicit sources:
• official
– documented rules
– documented business plans
– documented laws, regulations
– memos
• in-official
– artifacts (incl. documents)
– stories, anecdotes (location dependent)
Validity:
example: rules, regulations, laws
– intention
– date of installation
– friction with novel developments
– "officially still valid"
– overruled
Example: analysing history documents(Video “Kegworth fatal error” - training video HRA)
An example: the Kegworth accident
Failure of a safety critical system: a Boeing 737-400 crash at East Midlands
Hardware failure: left motor flutter
• up-scaling problem
• no operational test
Video document shows original material, and actual actors, but cockpit scenes are played by actors, based on cockpit voice recorder (another historic document)
Analysing history documents (continued)
A combination of problems merged after the original hardware failure:
• functionality of the control system
• representation of the system towards the users
• training and information
• organisation problems
• human failure
watch the video and try to identify problems in this task domain
5.2.4. Understanding implicit group knowledge Ethnographic methods
No a priory conceptual framework• bootstrapping - hanging around and identifying relevant concepts• informal representations - diary and journalist type records• activity theory, anthropology, sociology
Goal: understanding complex settings• multiple activities, multiple actors, multiple agendas• in natural groups
Focus on the situation, on work practice• knowledge and meaning are developed in “communities of practice”• knowledge is “ability to participate meaningfully”• people co-construct knowledge and skills, partly implicit or “tacit”• situated tasks, situated actions, situated understanding
Ethnographic methods (continued)
Ethnography is fieldwork
ethnographer is participant observer
• part of the situation, and co-constructor of events
• novice, if possible apprentice, or “legitimate peripheral participant”
• accepted to be “nosy”
the two viewpoints of ethnographic analysis
• the view of the analyst: being surprised, recording novel phenomena
• the view of the aboriginal: understanding in situation and the relevance of resources and representations
• diary needed from day one
Ethnographic methods (continued)
Ethnographic methods
• hanging around - identifying basic situational phenomena, getting accepted
• informal questions, triggered by activities and situation, based on growing understanding
• observation of routine and expert behaviour
• interaction analysis
Interaction analysis
Focus on interactions (humans, artifacts, situation)• identify "hot spots”
– documents, stories, participant observation• collect registrations - multiple instances• content log
– participants, situation, artifacts, technology, time– verbal and non verbal (body) language, territory, space, messages
Video review sessions:• comments of original participants• explanations, thinking aloud• ask to identify important issues• propose hypotheses
Interaction analysis (continued)
Interpretation:
• patterns in interaction, resources
• verifiable observations: records of different instances
• generalization: theory
• supplementary interviews, archive, observations
Group sessions:
• interdisciplinary
• replays to generate hypotheses
• max 5 minutes for discussion
• identify analytic categories
• confront hypotheses with other instances of hot spot theory
Generalizing ethnographic knowledge
Surprise index
• events and phenomena should be as predictable for the ethnographer as they are for the original actors
Systematic coverage:
• space, territory, off location
– make sure all relevant locations are considered (coffee lounge, corridors)
• temporal (shifts, seasonal changes, off hours)
– related events during commuting, during lunch hour
• actors (supporting people, spouses, unofficial consultants)
Generalizing ethnographic knowledge (continued)
Orientation - systematic viewpoints: choose one or more
• person oriented record– daily activities, interactions
• object oriented record– movements, change of ownership
• setting oriented record– at location, during shifts, movements in the space
• task / process oriented record– focus on process
Generalizing ethnographic knowledge (continued)
• event structure
– preparation, start, re-arrangements, intro of new objects
• temporal patterns
– day, season, technology driven rhythm
– “acted” continuity (receptionist, pilot, underground train “driver”)
• interaction structure
– verbal, instrumental, body language
– interrupted conversation (continuation triggered by availability of info or resources or time)
– just in time finalization triggered by interruption
Generalizing ethnographic knowledge (continued)
• spatial organization
– meaning of positions (head of the table)
– fixed places, free choice, stable or re-arrangements
• artifacts and documents
– location and movements
– construction and revision (documents)
– ownership (allowed activities related to objects)
– symbolic meaning (dress code, stethoscope)
– non physical objects (PIN, password, anecdotes, jokes)
ownership, trajectory, availability and inspectability
Example ethnographic record(video “Math Examination” VU 1994)
Task analysis for the design of an automated examination system (preparing questionnaires, registering for exam, taking the exam, scoring results, administering marks)
An examination: setting oriented record
• roles: professor, supervisors, students
• objects: forms, boxes, writing equipment, calculating equipment
• situation: time structure, entrance, scenes (hot spots)
watch video and identify and analyse unexpected events
5.3. GroupWare task analysis (GTA) Conceptual framework for collecting and modeling task knowledge
http://www.cs.vu.nl/~gerrit/gta
agents (with description of relevant attributes)
• actors (individual humans, institutes and groups, machines, systems)
• roles and their representation (incl. symbols)
• organization (structure of actors, roles, and allocation rules)
work
• task / goal structure / actions (acts, communication acts)
• decomposition and temporal structure
• protocols and (expert) strategies - partly situated
situation
• objects in object structure (type relations and semantic relations)
• environment
• history
GTA representations
conceptual entities
• template
structures
• tree
• flow
media registrations
• video
• sound track
• picture
Task World Ontology (Martijn van Welie)
Task Agent
Role
Event
Object
name(string)goal(string)start_condition(string)stop_condition(string)initial_state(string)final_state(string)duration(integer)frequency(string)type(enum)user_actions(string)system_operations(string)
name(string)skills(string) attitude(string)miscellaneous(string)
name(string)goal(string)name(string)
attribute(Name,Value)*action(Name)*
name(string)description(string)
Contains
Responsible
Performed_by
PlaysTriggers
Subtask
Uses
Triggers
Used_by
Subrole
Is
Performed_by