First Steps towards a Methodology for Unified Graph's Discrepancy Analysis

Researchers and IT professionals frequently use dataset comparison during software analysis. Additionally, they commonly make judgments based on discrepancies between two representations of the same item's set. To locate the error-prone areas of the system, developers may evaluate the densely linked regions of method call graphs in the context of their position in the package hierarchy tree. A universal technique for graphs, which can be utilized to unify the underlying process of discrepancy analysis, might help with these types of analyses. In this paper, we present a methodology for unified graph's discrepancy analysis, named Unigda. Its foundation is the previously established domain-specific discrepancy identification approach for cluster comparison. But to capture the similarity structures between the vertices of arbitrary graphs, we use several kinds of characteristic functions. 11Project no. TKP2021-NVA-09 has been implemented with the support provided by the Ministry of Innovation and Technology of Hungary from the National Research, Development and Innovation Fund, financed under the TKP2021-NVA funding scheme.