Metarule-guided association rule mining for program understanding

Abstract

Software systems are expected to change over their lifetime in order to remain useful. Understanding a software system that has undergone changes is often difficult due to unavailability of up-to-date documentation. Under these circumstances, source code is the only reliable means of information regarding the system. In this paper, we apply data mining, or more specifically, association rule mining, to the problem of software understanding i.e. given the source files of a software system, we use association rule mining to gain insight about the software. Our purpose is to explore the use of association rule mining for finding interesting associations within the software that can lead to program understanding. To make association rule mining more effective, we place constraints on the mining process in the form of metarules. Metarule-guided mining is carried out to find associations which can be used to identify recurring problems within software systems. We relate metarules to re-engineering patterns which present solutions to these problems. We apply association rule mining to five legacy systems and present results which show how extracted association rules can be helpful in analyzing the structure of a software system and in suggesting modifications to improve the structure. A comparison of the results obtained for the five systems also reveals legacy system characteristics, which can lead to understanding the nature of open source legacy software and its evolution.

Multi agent paradigm used to complexity measure for perfective software maintenance

“…In this work, we have divided the sole task of maintenance in different interacting sub-tasks and each is being assigned to a particular agent. The first agent is 'Coordinator agent' (COA), that receive the user requirement for change or enhancement and this agent send this requirement to the 'problem identification agent' (PIA) which identify the issue and revert back to the COA with problem analysis result and again the COA sends this information to the 'program comprehension agent' (PCA), this agent is having all program information such as file, classes, methods, variables and other valuable information that have been extracted from data mining methods [8] and stored in database. This PCA agent collects the information as per the requirement and reverts back to the COA.…”

Algorithm for mining association rules with multiple minimum supports based on FP‐Tree

“…Similarly, the second route can not identify the rules containing the infrequent data items and/or high frequency data items. Apparently, these routes are all relevant to the applications (Kona & Chakravarthy 2004;Maqbool et al 2005;ghodousian et al 2006). This paper introduces the idea that only one minsup is not enough to mine association rules in a database, because it can not discover the intrinsic natures and differences among data items appearing with different frequencies.…”