Detection Strategies Metrics-Based Rules for Detecting Design Flaws

1,210 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
1,210
On SlideShare
0
From Embeds
0
Number of Embeds
4
Actions
Shares
0
Downloads
14
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Detection Strategies Metrics-Based Rules for Detecting Design Flaws

  1. 1. Introduction Problem Demand Solution Implementation Evaluation Summary Detection Strategies Metrics-Based Rules for Detecting Design Flaws M.... N.....1 1 Universita della Svizzera Italiana, Switzerland Software Design and Evolution, WS 2009 Nowak Faculty of Informatics Detection Strategies
  2. 2. Introduction Problem Demand Solution Implementation Evaluation Summary Author Dr. Radu Marinescu Associate Professor - Department of Computer Science and Engineering "Politechnica" University at Timisoara Author of "Object-Oriented Metrics in Practice" Nowak Faculty of Informatics Detection Strategies
  3. 3. Introduction Problem Demand Solution Implementation Evaluation Summary Author Dr. Radu Marinescu Associate Professor - Department of Computer Science and Engineering "Politechnica" University at Timisoara Author of "Object-Oriented Metrics in Practice" Ph.D defense Mircea Lungu, Today, 17.30, A21, Red Building Nowak Faculty of Informatics Detection Strategies
  4. 4. Introduction Problem Demand Solution Implementation Evaluation Summary Outline 1 Introduction 2 Problem 3 Demand 4 Solution 5 Implementation 6 Evaluation 7 Summary Nowak Faculty of Informatics Detection Strategies
  5. 5. Introduction Problem Demand Solution Implementation Evaluation Summary Metrics Nowak Faculty of Informatics Detection Strategies
  6. 6. Introduction Problem Demand Solution Implementation Evaluation Summary Metrics Metrics ambiguous definitions noise relevance Nowak Faculty of Informatics Detection Strategies
  7. 7. Introduction Problem Demand Solution Implementation Evaluation Summary Metrics Metrics ambiguous definitions noise relevance Interpretation experience based no model showing symptoms not a disease Nowak Faculty of Informatics Detection Strategies
  8. 8. Introduction Problem Demand Solution Implementation Evaluation Summary Strategy Strategy "A detection strategy is the quantifiable expression of a rule by which design fragments that are conforming to that rule can be detected in the source code" Nowak Faculty of Informatics Detection Strategies
  9. 9. Introduction Problem Demand Solution Implementation Evaluation Summary Filters Semantical threshold value direction For example: Absolute: HigherThan, LowerThan Relative: TopValues, BottomValues Statistical direction For example: UpperQuantile, BelowMedian Nowak Faculty of Informatics Detection Strategies
  10. 10. Introduction Problem Demand Solution Implementation Evaluation Summary Strategy Choosing an appropriate filter 1 Absolute semantical filter 2 Relative semantical filter 3 Semantical filter with percentile values 4 Statistical filter Nowak Faculty of Informatics Detection Strategies
  11. 11. Introduction Problem Demand Solution Implementation Evaluation Summary Composition Logical Operators and, or, butnot filter Filtered Metric1 Set1 filter Filtered Composition Rules Final Metric2 Set2 Results filter Filtered Metric3 Set3 Nowak Faculty of Informatics Detection Strategies
  12. 12. Introduction Problem Demand Solution Implementation Evaluation Summary Metrics "God Class" syndrome. Weighted Method Count (WMC) Tight Class Cohesion (TCC) Access to Foreign Data (ATFD) (WMC(C), TopValues(25%))∧ (1) (ATFD(C), HigherThan(1))∧ (2) (TCC(C), BottomValues(25%)) (3) Nowak Faculty of Informatics Detection Strategies
  13. 13. Introduction Problem Demand Solution Implementation Evaluation Summary Process parsing Meta-Model Sources (Java, C++) Metrics using Detection Strategy Flaw List Detection Manual inspection Nowak Faculty of Informatics Detection Strategies
  14. 14. Introduction Problem Demand Solution Implementation Evaluation Summary Filter tuning 1 Experience and Literature 2 Reference Samples (Tuning Machine) 3 Evolution Analysis Nowak Faculty of Informatics Detection Strategies
  15. 15. Introduction Problem Demand Solution Implementation Evaluation Summary Case-study Version 1 93 KLOC, 18 Packages, 152 Classes, 1284 Methods Version 2 116 KLOC, 29 Packages, 387 Classes, 3446 Methods Evaluation methods Automatic Classification (differential between the versions) Manual Investigation (of the Version 1) Nowak Faculty of Informatics Detection Strategies
  16. 16. Introduction Problem Demand Solution Implementation Evaluation Summary Summary Results Automatic Classification accuracy over 50% with average over 67%. Manual Inspection method resulted in Accuracy of 87%. Nowak Faculty of Informatics Detection Strategies
  17. 17. Introduction Problem Demand Solution Implementation Evaluation Summary Summary Results Automatic Classification accuracy over 50% with average over 67%. Manual Inspection method resulted in Accuracy of 87%. Conclusion Method is very promising ! Nowak Faculty of Informatics Detection Strategies
  18. 18. Introduction Problem Demand Solution Implementation Evaluation Summary Related Work Quantification of Design Principles and Rules Using Correlations of Metrics for Design Inspections Nowak Faculty of Informatics Detection Strategies
  19. 19. Introduction Problem Demand Solution Implementation Evaluation Summary Discussion Questions and Discussion. Nowak Faculty of Informatics Detection Strategies

×