AUTOMATED EXTRACTION OF RULES EMBEDDED IN SOFTWARE APPLICATION CODE USING MACHINE LEARNING

摘要

Legacy codes of software applications are required to be modernized andmigratedto the latest technology. Migration of legacy codes requires extraction ofhidden rulescomprised in the application code and translating them to meaningful outputwhich iscumbersome. Thus an intelligence in the entire rule extraction and translationprocess isrequired for reducing the complexity and risk. The present disclosure providesautomatedextraction of rules embedded in software application code using machinelearningtechnique(s) (MLT). In the present disclosure, rules embedded in the softwareapplicationsource codes are extracted based on a control flow and data flow analysis.Further, theextracted rules are translated into a target defined format based on mappingof parametersassociated with extracted rules with a pre-stored meta data wherein the mappedparametersare classified into one or more categories using the MLT. The translated rulesare analyzedto obtain a validated set of rules.