160 likes | 300 Views
Case Based Reasoning Project Presentation. Presenter: Madan Bharadwaj Instructor: Dr. Avelino Gonzalez. Agenda. Problem Definition Approaches considered Program Structure Relevant Results Conclusion Summary. Introduction. What is Case Based Reasoning? What is knowledge in CBR?
E N D
Case Based Reasoning Project Presentation Presenter: Madan Bharadwaj Instructor: Dr. Avelino Gonzalez
Agenda • Problem Definition • Approaches considered • Program Structure • Relevant Results • Conclusion • Summary
Introduction • What is Case Based Reasoning? • What is knowledge in CBR? • Implementation issues in CBR
Problem Definition • Given:# Case Library of 21 cases# 7 test cases# Information about the domain • Develop a CBR System • Using any Programming Language
Approaches Considered • Use Domain Thresholds from Handouts • Extract rules from Case Library • Pattern Matching and Thresholds • Pattern Matching
Chosen Approach • Pattern Matching only • Justification
Choice of Prog. Language • Visual Basic vs C++ • VB advantages • C++ Disadvantages • Pros of using C++ • Ideally…
Program Structure • Initialize Case Library • Take Test Case Input • Pattern Matching with all cases in Case Library • Compare Pattern Matching results for closest match • Display
Results • To Prove Credibility of System test with # Library Case# Sample case similar to Library Case • Test with given test cases
Library Case Results • Chosen Library Case: Case #2 • 100% Match • Closest Match: Library Case #2 • Diagnosis: LEAK
Test case (like Lib. Case) • Input parameters similar to Library Case #2 • 95.3639 % Match • Closest Match: Case #2 • Diagnosis: LEAK
Other Test Case Results • Answers unknown • Checked intuitively for correctness
Adaptation • Adaptation not used • Possible, but needs expert help during design time to ensure correctness
Augmentations • Graphical Display • Case Library in Database • More checks for faulty inputs
Conclusion • Case Library dependent • Better performance in Weak-theory domains • Simple Design • Execution time • Lack of Intuitive Knowledge
Summary • Purely Case Based Approach • Used Pattern Matching • Tested with fabricated cases • Augmentations • Conclusions derived