Modeling decision tables

1
Prologa

• Modeling decision tables

2
Prologa
3
LIRIS Leuven Institute for Research on
Information Systems
4
Prologa (PROcedural Logic Analyzer)

• Computer supported
• Design
• Validation
• Manipulation
• Use, and
• Transformation
• of Decision tables

5
Importance and applications

• 1. Conditional logic
• systems analysis
• business rules
• software engineering
• procedures, texts, regulations
• test case specifications

6
Importance and applications

• 2. Knowledge-based systems
• Knowledge acquisition
• Validation and verification of knowledge systems
• (consistency, completeness, correctness)
• Implementation of decision logic

7
PROcedural Logic Analyzer
8
Emphasis on

• Interactive specification of decision logic
• User friendly and automatic manipulations
• Automatic verification and validation
• Interfaces to various representations
• (trees, code, rules)
• Optimizations

9
An integrated development approach
10
Modeling

• Research topics
• a) Knowledge base decomposition (modularisation)
• clustering, factoring
• b) Filling incomplete knowledge
• Induction algorithms (ID3),
• Neural networks
• c) Visual Modeling
• d) Fuzzy DTs
• Fuzzy conditions, actions, implications
• Fuzzy decision making
• Fuzzy VV
• e) Temporal aspects
• f) Constructed negation

11
Optimization

• Research topics
• a) Optimal table contraction (rule reduction)
• Contraction min columns (given condition
  order)
• Optimal order min columns
• (with ordering constraints)
• b) Optimal test sequences (decision tree)
• (given condition test times and column
  frequencies).
• c) Minimization of rules per action
• rewriting specifications
• d) Comparison of decision tree induction with
  the generation of optimal execution trees
• e) Developing heuristic algorithms

12
Implementation

• Research topics
• a) The choice trees, rules, ...
• b) KBS performance
• c) Integration with OO
• d) Generating applications
• e) Descriptive theories of the relationship
  between visual querying and decision making
  performance

13

• Modeling

Basic Modeling Process
14
Basic modeling process

• Start new table
• Input conditions and actions
• Input decision rules
• Display table
• Optimize table

15
Start new table
16
Basic modeling process

• Start new table
• Input conditions and actions
• Input decision rules
• Display table
• Optimize table

17
Conditions and actions

• Detecting decision criteria, relevant
• states and possible outcomes

18
Action and condition input
19
Basic modeling process

• Start new table
• Input conditions and actions
• Input decision rules
• Display table
• Optimize table

20
Decision rules

• Relations between decision criteria and actions
  to be be executed
• General rules, exceptions, ...
• Based upon starting text, interview, examples,
  etc.

21
Decision rules

• Action(s) IF Condition combination(s)
• Various refinements
• Generally if
• If and only if
• Always if
• Only possible if
• Impossible if
• ...

22
Decision rule input
23
Basic modeling process

• Start new table
• Input conditions and actions
• Input decision rules
• Display table
• Optimize table

24
Table display
25
Expanded table
26
Verification Validation

• Automatic checking for
• Completeness
• Consistency
• Correctness
• Redundant conditions
• Redundant actions
• Redundant combinations

27
Avoiding contradictions
28
Verification report
29
Various manipulations

• Add, update, delete, reorder conditions and
  actions
• gt immediately reflected in table
• Table layout

30
Basic modeling process

• Start new table
• Input conditions and actions
• Input decision rules
• Display table
• Optimize table

31
Various optimizations

• Layout
• Contraction
• Optimal condition order
• Optimal execution time
• Optimal rules

32
Contracted decision table
33

• Modeling

Additional features
34
Additional features

• Fill by mouse
• Table manipulation
• Automatic verification
• Optimal contraction
• Action and condition dictionary
• Table structures
• Interface features
• Consultation of tables

35
Fill by mouse
36
Fuzzy tables
37
Import from MS Excel
38
Import from MS Excel
39
Decision Tree
40
Export options
41
Pascal generation
42
COBOL generation
43
AionDS table generation
44
Optimal execution code
45
Minimal rules
46

• Modeling

Structures of decision tables
47
Structures of decision tables
48
Decision table modularization
49
Verification between tables
50

• Consultation of the knowledge

Consultation environment
51

• Tables
• (logic)

Text, explanation multimedia
Consultation mechanism
Application
52
Filling in criteria
53
Result of the consultation
54

• ...