PERFORMANCE ANALYSIS OF HIERARCHICAL INSTRUMENTATION MINING PREDICATED BUG SIGNATURE METHOD FOR BUG LOCALIZATION IN PROGRAMS

• Main Author: Rusrin Pratama, Nando
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/66314
• Institution: Institut Teknologi Bandung

The purpose of this research is to analyze the performance of the hierarchical instrumentation mining predicated bug signature (HIMPS) method in localizing bugs in the program. The HIMPS method is a bug localization method that generate bug signatures from predicates in the program and uses hierarchical instrumentation (HI) techniques to instruments input program into a list of predicates. The HI technique is a technique for performing instrumentation process of input program in phases so the instrumentation can be carried out selectively and optimally. Based on the HI technique, the instrumentation process can be divided into 2 phases, that is the coarse-grained phase and fine-grained phase. The coarse-grained phase is the phase where the process is at the functional level, and fine-grained phase is the phase where the process is at the predicate level. The purpose of this research is to analyze and examine the effect of using HI techniques on the performance of the HIMPS method in terms of efficiency and effectiveness. In terms of efficiency, the parameters used are time and memory, and in terms of effectiveness, the parameters used are accuracy. The performance analysis of the HIMPS method is carried out with a conceptual study and validated through experiments. The conceptual study was carried out by conceptual complexity analysis with big-O notation and conceptual quality analysis at the discriminative significance value. From the conceptual study, it can be concluded that the variables in the input program that influence the performance of the HIMPS method are the number of functions, the number of predicates, and the number of test cases. The conclusion of this research is the performance analysis of the HIMPS method in the conceptual study was proven correct and validated based on experiments. From the analysis of the HIMPS performance and the analysis of experimental results, it can be concluded that the performance of the HIMPS method is carried out by the use of HI techniques, both in terms of efficiency and effectiveness. The variables that influence the performance of the HIMPS method are the number of functions, the number of predicates, and the number of test cases. In terms of efficiency, after simplification it can be concluded that the influence of the variable number of functions is O(1) or constant, the number of predicates is O(n) or linear, and the number of test cases is O(n) or linear. In terms of effectiveness, after simplification it can be concluded that the influence of the variable number of functions is O(1) or constant, the number of predicates is O(n) or linear, and the number of test cases is O(nm ) or exponential.