Estimation of Program Reverse Semantic Traceability Influence at Program Reliability with Assistance of Object-Oriented Metrics

1
Estimation of Program Reverse Semantic
Traceability Influence at Program Reliability
with Assistance of Object-Oriented Metrics
V.Rogalchuk, A.Tyrva, A.Khomonenko Petersburg
State Transport University
2
Reverse Semantic Traceability

• Maintenance and refactoring expense reduction
• Reasonable to use when development team is
  changed
• ? Influence on cost (COCOMO II) and reliability

3
Object-oriented metrics

• Chidamber Kemerer metrics suite (response for a
  class, depth of inheritance tree, number of
  children, number of methods and attributes etc.)
• Possible to gather metrics values from UML
  diagrams automatically

4
Reliability forecast

• Logistic regression

xi i-th metric value, ßi i-th metric ratio.
Model parameters ßi are calculated using
maximum likelihood method
5
Reliability forecast

• Example fault possibility for child-class
  (DIT1) with three attributes (NAI3) under
  condition that there are two other classes with
  method parameter types of this class (OCMEC2)

6
Estimation of Reverse Traceability Influence at
Program Reliability

• Correction of ß0 ratio is required to consider
  this influence

7
Estimation of Reverse Traceability Influence at
Program Reliability
Fault type Fault type percentage PE, Reverse traceability influence on fault detection IT, Fault detection percentage ?PE,
Error handling faults 2 50 1
Unoptimized Code 3 50 1,5
Incorrect Logic in Code 12 75 9
Scheduling Error 7 10 0,7
Static Data Error 22 10 2,2
Shell Script Error 14 10 1,4
Database Error 7 25 1,75
Environment Error 25 50 12,5
Data File Error 5 80 4
Incorrect Mapping 3 50 1,5
Total 100 35,55
8
Estimation of Reverse Traceability Influence at
Program Reliability

• Fault possibility with an allowance for reverse
  traceability
• 0.5422 0.35550.5422 0. 5422 0.1928 0.3494

9

• ?