Case-based reasoning

1
Case-based reasoning
2
What is case-based reasoning?

• An approach to building KBS which is radically
  different to the rule-based and other
  knowledge-representation approaches we have seen
  so far.
• The principle is to find a solution which has
  been shown to solve problems like your current
  problem in the past, and adapt it so that it
  solves the current problem.

3
What is case-based reasoning?

• This has a certain psychological plausibility as
  a model of what the expert-decision-maker
  actually does when solving a problem.
• Based on research by Riesbeck Schank (1989). A
  good comprehensive description is to be found in
  Kolodner (1993).

4
What is case-based reasoning?

• Three quotes from Roger Schank
• "Humans use cases because they don't know what
  they know - they don't know their own rules -
  they do things non-reflectively."
• "The key process in intelligence is the
  reminding process".
• "People don't ever reason from first principles.
  They always choose a matching case. It may be a
  bad match, but in that case they need more
  experience.

5
How a CBR system works the knowledgebase

• The knowledge base contains a collection of
  representative cases (of faults, say, if the
  system is concerned with fault diagnosis), with
  their
• symptoms,
• causes,
• and treatments.

6
How a CBR system worksthe process

• The user is instructed to provide the (relevant)
  features of the current case.
• The similarity between this set of features, and
  the features characteristic of each of the stored
  cases is calculated, and the best match is chosen.

7
How a CBR system worksthe process
1 3 4 7 5 2
1 3 5 7 5 3
1 3 4 7 4 4
2 8 4 7 4 3
1 4 4 7 0 2
Case 1
Case 2
Case 3
Case 4
Case 5
8
How a CBR system worksthe process
1 3 4 7 5 2
1 3 5 7 5 3
1 3 4 7 4 4
2 8 4 7 4 3
1 4 4 7 0 2
Case 1
Case 2
Case 3
Case 4
Case 5
9
How a CBR system worksthe process
1 3 4 7 5 2
1 3 5 7 5 3
1 3 4 7 4 4
2 8 4 7 4 3
1 4 4 7 0 2
Case 1
Case 2
Case 3
Case 4
Case 5
10
How a CBR system worksthe process
1 3 4 7 5 2
1 3 5 7 5 3
1 3 4 7 4 4
2 8 4 7 4 3
1 4 4 7 0 2
Case 1
Case 2
Case 3
Case 4
Case 5
11
How a CBR system worksthe process
1 3 4 7 5 2
1 3 5 7 5 3
1 3 4 7 4 4
2 8 4 7 4 3
1 4 4 7 0 2
Case 1
Case 2
Case 3
Case 4
Case 5
12
How a CBR system worksthe process
1 3 4 7 5 2
1 3 5 7 5 3
1 3 4 7 4 4
2 8 4 7 4 3
1 4 4 7 0 2
Case 1
Case 2
Case 3
Case 4
Case 5
13
How a CBR system worksthe process

• The features which have been identified as
  important in the stored cases, and which the user
  is asked about, are known as indices.
• Each has a value. In the example I just showed
  you, each was represented by a number.

14
How a CBR system worksthe process

• If necessary, this case is adapted so that it is
  a better match for the current circumstances.
• The case is then presented as the solution, with
  the opportunity to examine the 'precedent' case.

15
How a CBR system works

• The sequence of operations, for a
• simple CBR system1) assign indices2) retrieve
  a similar case

16
Flow chart for a simple CBR system
Input
1. Assign indices
2. Retrieve
Output
17
How a CBR system works

• The sequence of operations, for a
• full-blown CBR system1) assign indices2)
  retrieve a similar case3) modify the past
  case4) test the case 5a) assign indices to this
  new case, and store as a working solutionOR
• 5b) explain failure, repair the solution, and
  test again.

18
Flow chart for a full-blown CBR system
Input
1. Assign indices
2. Retrieve
3. Modify
5b. Store
4. Test
5a. Assign indices
6b. Repair
Failed solution
Working solution
6a. Explain
19
Available techniques for case memory organisation

• Memory organisation by
• linear ("flat") case memory
• case hierarchy
• nested cases
• decision-tree orientated memory
• knowledge-guided indexing

20
Available techniques for case retrieval

• Retrieval by
• Nearest neighbour case matching
• Weighted nearest neighbour case matching
• Decision tree methods
• Knowledge-guided retrieval

21

• The last four memory organisation approaches, and
  the last two retrieval approaches, might be
  thought of as hybrid systems.

22
Nearest neighbour algorithm an example

• Suppose that we have a sick soyabean plant, and
  we wish to discover which of a number of known
  specimens of sick soyabean plants it is most like.

23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
Nearest neighbour algorithm an example

• Choose (lets say) three characteristics of the
  leaves that can be represented as numbers
• Amount of the leaf that is covered by the
  discolouration
• Lightness of the discoloured parts of the leaf
• Lightness of the remaining parts of the leaf.

27
Nearest neighbour algorithm an example

• Suppose that the first two cases to be matched
  are
• case 1 coverage - 8
• lightness-1 - 4
• lightness-2 - 6
• case 2 coverage - 10
• lightness-1 - 7
• lightness-1 - 6

28
Nearest neighbour algorithm

• This can be treated as two points in
  three-dimensional space
• x, y, z coordinates of case 1 (8, 4, 6)
• x, y, z coordinates of case 2 (10, 7, 6)

29
Nearest neighbour algorithm
y
10 9 8 7 6 5 4 3 2 1
x
1
2
3
4
5 6
7 8 9 10
0
1 2 3 4 5 6 7 8 9 10
A system of 3-dimensional co-ordinates
z
30
Nearest neighbour algorithm
y
10 9 8 7 6 5 4 3 2 1
? - case 1
x
?
1
2
3
4
5 6
7 8 9 10
0
1 2 3 4 5 6 7 8 9 10
The 1st case represented as a point
z
31
Nearest neighbour algorithm
y
10 9 8 7 6 5 4 3 2 1
?
? - case 2
x
1
2
3
4
5 6
7 8 9 10
0
1 2 3 4 5 6 7 8 9 10
The 2nd case represented as a point
z
32
Nearest neighbour algorithm
y
10 9 8 7 6 5 4 3 2 1
? - case 1
?
? - case 2
x
?
1
2
3
4
5 6
7 8 9 10
0
1 2 3 4 5 6 7 8 9 10
The two cases represented as points
z
33
Nearest neighbour algorithm
y
10 9 8 7 6 5 4 3 2 1
? - case 1
?
? - case 2
x
?
1
2
3
4
5 6
7 8 9 10
0
1 2 3 4 5 6 7 8 9 10
The distance between the two cases
z
34
Nearest neighbour algorithm
y
10 9 8 7 6 5 4 3 2 1
? - case 1
? - case 2
? - case 3
?
x
?
1
2
3
4
5 6
7 8 9 10
0
1 2 3 4 5 6 7 8 9 10
Adding a third case (2, 3, 9)
?
z
35
Nearest neighbour algorithm

• There is a simple formula that tells you the
  distance between two points in 3-dimensional
  space.
• To find out whether case 1 is more similar to
  case 2 or to case 3, you simply calculate the two
  distances, and pick the smaller of the two.

36
Nearest neighbour algorithm

• To find out which of a whole series of cases case
  1 is most similar to, calculate the distance from
  case 1 to each of them, and pick the smallest
  figure.

37
Nearest neighbour algorithm

• Suppose it was 4 features, or 7, or 100? Would
  you have to draw 4-dimensional or 7-dimensional
  or 100-dimensional graphs?
• No, its simply necessary to have a formula for
  calculating distances in 4, or 7, or
  100-dimensional space, and such formulae are
  readily available.

38
Case adaptation

• "Fixing" inconsistencies between diagnosis and
  symptoms.
• Techniques
• the end user does it
• knowledge-based (qualitative reasoning, etc)
• a fixed procedure.

39
Case adaptation

• Note that there is a problem about updating the
  case-base with adapted cases.
• Since the new case isnt exactly like any of the
  cases in the case-base, it cant really be said
  to have been solved by the expert judgement that
  was used to build the case-base in the first
  place.
• There is a real chance that the conclusion that
  the system came to is wrong in this case
• If wrongly concluded cases are added to the
  case-base, it becomes progressively degraded.

40
Case adaptation

• Typically, the procedure is to put fresh cases
  into a special file, and have the Domain Expert
  pass judgement on them before they are added to
  the case-base.

41
Appropriate domains

• CBR is suitable
• when the domain is broad but shallow.
• when experience rather than theory is the primary
  source of information.
• when the requirement is for the best available
  solution, rather than a guaranteed exact
  solution.
• when solutions are reusable, rather than unique
  to each situation.

42
Example of a successful system

• CBR is particularly used for help-desk
  applications.
• For instance the COMPAQ SMART system.

43
Example of a successful system

• The problem was that
• Thousands of customers were calling Compaq
  directly every day, requesting support.
• Many of the staff were new there was a major
  training problem.
• There was a need for consistent accurate
  answers and responses
• There was a need for retention of corporate
  knowledge.

44
Example of a successful system

• The COMPAQ SMART system, once developed and
  installed, succeeded in solving 85-95 of calls.
• Typical time to solve a problem was less than 2
  minutes.

45
Advantages of CBR

• Case-based reasoning
• tends to focus on the problem's essential
  features.
• can solve problems in domains that are only
  partially understood.
• can provide solutions when no algorithmic method
  is available.
• can interpret open-ended and ill-defined
  concepts.

46
Steps in building a case-based reasoning system

• 1. Obtain data for cases.
• 2. Design cases based on data.
• 3. Determine the case memory structure.
• 4. Decide the case retrieval method.
• 5. Decide whether a case adaptation procedure is
  appropriate (and, if so, implement it).
• 6. Develop the rest of the system (e.g. the user
  interface).

47
Some currently-available CBR tools (with vendors)

• Esteem (Esteem Software)
• CBR Express CBR v.2.0 (Inference)
• ReMind (Intelligent Applications, Cognitive
  Systems)
• ReCall (ISoft)
• KATE-CBR (Acknosoft)
• Some of these are UK products, some American,
  some French.

48
Example of a large CBR project the Cassipoée
system

• Used a combination of inductive and CBR
  techniques.
• Written using KATE-CBR, by AcknoSoft of Paris, on
  behalf of an engineering firm owned by General
  Electric and SNECMA.
• A diagnostic system for aircraft engines CFM
  56-3 engines in Boeing 737s and Airbus A340s.

49
Example of a large CBR project the Cassipoée
system

• The cases came from a legacy database of 23000
  engine maintenance reports, built up over 8
  years.
• Experienced engineers worked over the cases,
  eliminating items where there was no diagnosis or
  mis-diagnosis, and duplicates.
• This left 16000 cases, each with up to 100
  features.

50
Example of a large CBR project the Cassipoée
system

• Case selection was by a decision tree, generated
  from the cases.
• This directed the questioning of the user, to
  provide a set of symptoms, to select cases.

51
Example of a large CBR project the Cassipoée
system

• Extra features
• Integrated with an Illustrated Part Catalogue
• Generates reports of reliability and
  maintainability using EXCEL
• Uses e-mail to collect maintenance reports
  world-wide.

52
Example of a large CBR project the Cassipoée
system

• Success
• Very fast diagnosis reduced diagnosis time by
  50
• Won 1st prize for innovative software
  applications at the European XPS show, Germany,
  March 1995.

53
A note on knowledge acquisition

• In rule-based reasoning, knowledge is extracted
  from experts and encoded in rules. This is often
  difficult to do. In case-based reasoning, most
  (but not all) knowledge is in the form of cases. ?

54
A note on knowledge acquisition

• Case-based reasoners also need the same semantic
  knowledge that rule-based reasoners need. In
  addition, case-based reasoners need adaptation
  rules and similarity metrics - more types of
  knowledge, but perhaps knowledge that is easier
  to acquire. ?

55
A note on knowledge acquisition

• Several recent studies point to the relative
  ease with which case-based reasoners can be built
  as compared to building the same rule-based
  systems. ?
• Kolodner (1993), p.94

56
Knowledge acquisition

• In one study, the Digital Equipment Corporation
  commissioned two systems (for customer technical
  support), with equivalent functionality.
• One, called CANASTA, was rule-based one, called
  CASCADE, was case-based.

57
Knowledge acquisition

• CANASTA took 960 days of development time
• CASCADE required 105 days.
• However, the personnel required for the CANASTA
  development were more valuable than those
  required for CASCADE
• if one takes account of this, the development of
  CANASTA took the equivalent of 1600 days, and
  CASCADE the equivalent of 193 days.

58
Knowledge acquisition

• CANASTA took 960 days of development time
• CASCADE required 105 days.
• However, the personnel required for the CANASTA
  development were more valuable than those
  required for CASCADE if one takes account of
  this, the development of CANASTA took the
  equivalent of 1600 days, and CASCADE the
  equivalent of 193 days.

59
Knowledge acquisition

• The accuracy and efficiency of the two systems
  were reckoned to be equivalent.
• The continuing maintenance costs of CANASTA were
  high, while those of CASCADE were negligible.
  (Simoudis, 1991 1992).

60
A comparison between rule-based case-based
reasoning
61
A comparison between rule-based case-based
reasoning
62
A comparison between rule-based case-based
reasoning
63
A comparison between rule-based case-based
reasoning
Advantages
64
A comparison between rule-based case-based
reasoning
Disadvantages
65
(No Transcript)