Kinds of System and Problem Frames

1
Kinds of System and Problem Frames

• Lecture 6

2
Getting a Grip on Problems

• Are use cases enough?
• Two problems
• The Mayday system
• The Volunteer system
• Common kinds of system
• repository
• simulation
• Matching systems
• Review

3
Getting Started

• How do you start to tackle a new systems
  analysis/design problem?
• Use a Human-centred / organisationally-centred
  approach to problem identification
• Do detailed analysis - 100s of use cases,
  sequence diagrams, class diagrams
• But whats the big picture? What does the system
  DO?

4
Problem

• An organisation has a mission to put volunteers
  with certain skills, interests and available time
  in touch with community groups or individuals
  with specific needs.
• Assignments last for a period of time ranging
  from a day to a year. Volunteers can often take
  several assignments at a time, and groups may
  need several helpers. The organisation tries to
  achieve a good match between volunteers and
  groups, resulting in valued, lasting
  relationships
• Currently assignments arrive at the rate of 20 a
  day, and new volunteers arrive at the rate of 5 a
  day.

5
What kind of System is it?

• system is a repository
• system is a simulation of the real world
• system is a resource manager
• system is a match-maker
• system is a controller of real world activity
• system is a co-ordinator of human activity
• system is a tool for construction of an artifact
• system is a learning system
• .....
• Real systems have aspects of several kinds but it
  helps to focus of each viewpoint in turn

6
System is a Repository

• We can view many systems as simple repositories
  of data.
• Data is collected, stored, kept safe from falling
  into the wrong hands, from being lost.
• Data can be extracted, reorganised, removed if no
  longer relevant
• Data should only be accepted if it conforms to
  rules for good data -integrity rules
• Simple-minded Database view of systems - the CRUD
  view.

7
Repository

• In the problem, we need to be able to store data
  about volunteers and clients
• To be able to add new data at a rate to keep up
  with arrival of new volunteers and clients
• Need to address such issues as how to keep the
  data secure, available, safe
• The Scope of this view is the Computer System
• Metaphor SYSTEM IS A BANK

8
System is a Simulation

• System simulates some aspects of the real world
• Design problems
• how much of the real world to model, in what
  detail to support needs of users, now and in the
  future?
• how to ensure that the model and the real world
  stay closely in synchrony - that changes to the
  real world (a volunteer moves house) are
  reflected in the model (address field updated)
  quickly and accurately
• how to detect changes in the real world and
  communicate these to the system
• how and what derived information is needed

9
Simulation

• The scope of this view is the whole Information
  System
• Jackson calls this the Information Display
  Frame
• Metaphor SYSTEM IS A MAP
• System is a Simulation has many subtypes
• system is a resource manager
• system is a match maker
• system is a classifier
• ....

10
System is a Resource Manager

• Allocation of scarce resources, such as
• seats on flights
• rooms for lectures
• beds for patients
• Problem is to maximise utilisation of resources
  whilst minimising delay and loss of business
• A common sub-problem is to match a clients needs
  to available resources

11
System is a Match-Maker

• Many systems have a matching task at their core.
• fingerprint matching
• patient/donor matching for transplant surgery
• patients to clinical trials
• interns to placements in hospitals
• parents with similarly challenged children
• DNA samples
• electronic devices - speakers to amplifiers
• search request to relevant documents
• incoming news items to information subscribers
• finding a rescue boat
• number plate recognition in Londons Congestion
  Congestion Charging System

12
System is a Classifier

• Classification separates data into two or more
  classes
• Filtering Email Spam
• Classifying students by grade of degree
• Classifying credit card transactions as
  fraudulent
• Detecting spelling mistakes
• METAPHOR SYSTEM IS A SIEVE

13
Classification Errors
Good mail
Spam
accept
correct
false positive
false negative
correct
reject
The two errors are in conflict - we can decrease
the risk of a false positive (reject more Spam)
but we increase the risk of false negatives
(rejecting good email) a TRADE-OFF.
14
Matching in detail

• Matching task typically involve
• two sets of individuals e.g.
• the volunteers / available rescue craft -
  Resource
• the community tasks / the distress vessel -
  Requirement
• adequate representations of both
• a fitness function which calculates how well
  matched a Requirement is to a Resource
• Different allocation constraints
• single allocation - can only rescue one boat at a
  time
• batch allocation - best solution for a set of
  volunteers
• no allocation - searching the web

15
Single Allocation

• Allocation to a single Requirement
• long list the Resources - eliminate the
  obviously unsuitable (classification)
• compute fitness between Requirement and each
  remaining Resource
• rank the Resources in fitness order for a short
  list
• ? user selection from short list on basis of
  additional information unknown to system

16
Creating a Fitness Function

• Mayday
• can use minimum time to reach position, but a
  more distant boat may have better rescue, medical
  support, more fuel, hence need for judgement
• Community volunteers
• travel distance between volunteer and group may
  be one factor (complicated by roads and bus
  routes)
• other criteria will be categorical - skills,
  likes and dislikes, age
• is electrician closer to mechanic than builder?
• hard to combine into a single, meaningful measure
  of fitness - combining travel distance and skill
  match

17
Fuzzy matching

• How close are two ages - e.g. in a dating agency,
  age of applicant, desired age of partner?

1.0
?
0.0
min
max
age
18
Combining fuzzy values

• Multiple qualities(dimensions) need to be
  compared but how?
• In the Mayday system, we combine two dimensions
  (Lat and Long) using a variant of Pythagoras
  (X2Y2 R2 )
• In the Volunteer system
• put weights (reflecting importance) on each
  dimension ( age, travel, skill..)
• compute sum of squares of weighted differences

19
System is a learning system

• To define the Fitness criteria
• need to carefully research the criteria for a
  good match, but not likely to get measures,
  weights right first time
• Good fitness functions need to be learnt
• collect evidence of the quality of the resultant
  match in order to refine matching criteria, to
  detect missing criteria, to determine the weights
  to be given to separate criteria
• Learning affects the total system
• of data collected, of recruitment of volunteers,
  as well as amendment of the software

20
A Learning system

• Learning cycle
• action - put current rules into practice
• observation - record results of rules
• analysis - analyse reasons for any differences
• change - modify the rules to improve performance
• Feedback loop, comparing goals with actual
  performance

action
observation
change
analysis
21
Framework Approach to Analysis

• Identify core task areas
• Apply common framework to specific case
• Framework provides
• standard terminology
• checklist of issues which need to be addressed
• index of solutions to commonly occurring problems

22
Commercial Of the Shelf Software (COTS)

• Software exists for most business needs
• payroll
• order processing
• general ledger
• human resources
• e-commerce
• e.g. SAP, SAGE ..
• but analysts need to match business needs to COTS
  capability, and customise generic software for
  local business rules.

23
Related Material

• Michael Jackson (2001) Problem Frames,
  Addison-Wesley takes a more formal approach.

24
Tutorial

• Choose a couple of different matching problems,
  or identify your own.
• Research the fitness function which would be
  appropriate for these problems
• Identify the processes required to implement a
  learning system around each matching system