Case-Based Reasoning 690. 19E-konul Nabiyeva Knowledge Engineering- letafet Ghardashova

Case-Based Reasoning 690.19E-Konul Nabiyeva Knowledge Engineering- Letafet Ghardashova

What is Case?

What is Case?

A case has two parts:

1.Description of a problem or a set of problems

2.Description of the solution of this problem.

Possible additional components might be explanations, and comments on the quality of the solution, etc. Cases represent experiences, i.e., they record how a problem was solved in the past

What are CBR Systems?

In CBR, information is stored in a KB in the form of cases, rather than rules. When a new situation is encountered, the CBR system reviews the cases to try to find a match for this particular situation. If

a match is found, then that case can be used to solve the new problem .Case-based reasoning works in a similar way to which humans select a course of action from experience

Everyday examples of CBR

• An auto mechanic who fixes an engine by recalling another car that exhibited similar symptoms
• A lawyer who advocates a particular outcome in a trial based on legal precedents or a judge who creates case law.
• An engineer copying working elements of nature (practicing biomimicry), is treating nature as a database of solutions to problems.

Steps in CBR

Collect the important features that define each new case presented to the system.
Retrieve past cases matching these features most closely.
Select the cases most relevant to the current problem.
Where necessary, adapt the stored cases to solve the current problem.
Validate the new solution and store as a new case.
Where a case match is not found, then find an alternative solution and record both problem and solution.

The classical CBR algorithm

Retrieve the most relevant cases for
comparison with the current problem.
Reuse knowledge from those
cases which help in solving
the current problem
Revise the proposed solution,
using the case information.
Retain details of the current problem as a new case for future reference.

Case-Based Reasoning Algorithm

It is suggested to use Case-Based Reasoning (CBR) method to solve a task on adapting decision to experience related to solving similar tasks. The process of searching for information required is a subsequent implementation of following actions: obtaining (retrieving) a precedent, searching for the closest decision in Case Base (CB) considering similarity degree on criteria chosen, adjustment and saving, decision making.

CBR Examples

In the following example, we will look at how a case base containing only two cases deals with a problem with a printer. The storage of the cases in the case base might be represented by Figure 2.12. Notice that the solution records not only the diagnosis of the cause of the problem but also the action taken

CBR Examples

Figure 2.13 shows the two cases in the case base together for comparison. Each case describes one particular situation and all cases are independent of each other. When a new problem has to be solved, several observations of the current situation are made. These observations define the new problem. Not all feature values need to be known. Note that the new problem is ‘case’ without solution component.

When Are Two Cases Similar?

The new problem is compared with each case in the case base and the most similar case is selected (Figure 2.15).

Cases are ranked according to their ‘similarity’ and therefore similarity is the most important concept in CBR. We can assess similarity based on the similarity of each feature and similarity of each feature depends on the feature value.

The degree of similarity can be

expressed by a real number between

0 (not similar) and 1 (identical).

These can be assigned to particular

features based on experience.

Similarity Computation by Weighted Average

However, the importance of different

features may be different and there

is therefore the need to apply weights to features. High importance: Problem,

Paper empty, Printer ready.

Low importance: Printer type.

Note that the 1/25 component is
arrived at from 1/total feature
weights, 1/(6 + 1 + 6 + 6 + 6).

Exercise

• Sketch a diagram relating the problem to Case 2

2. Conduct a similarity computation for Case 2

3. Determine which of the cases is more similar to the problem.

Case 1 is therefore more similar—thus the diagnosis ‘Paper jam’ would be made. If the diagnosis is incorrect then a technician must carefully work out what is wrong with the printer. Having done this the new case (problem plus diagnosis and solution) can be stored in the case base and our case base thus contains another example of problem and solution

Pros/Cons

• Reduction in the amount of knowledge acquisition actually needed, because the CBR system searches current cases for solutions rather than inferring solutions from a rule base.
• The CBR system learns over time by adding new cases to the KB.

• Storing of cases in the KB. Care is needed to ensure that cases are referenced correctly with appropriate attributes.
• Not providing good presentation of information to the user

Advantages

Disadvantages

Resources

• https://www.slideshare.net/theintegralworm/case-based-model-based-reasoning
• Introduction to knowledge engineering S.Kendall,M.Creen
• https://www.slideserve.com/axelle/case-based-reasoning

THANKS