Compliance-Evaluation-of-Database-Application-Systems

The mismatches between the implemented database application systems and its data specification standard description files significantly affects the accuracy of enterprises' estimation of data assets based on data standard files. In this study, we proposed an automated approach for evaluation and discovery of these mismatches, greatly reducing the cost of manual evaluation. We frame the mismatch discovery problem as an entity matching computation on two distinct graphs, respectively constructed from the database of application systems and its data specification standard. A new entity alignment method was proposed. It leverages a common view to model the global structure information of the entities, meanwhile, it models the relationship information around the entities with the specific view, and then integrates them into the entity representation to estimate matching similarity and filter outlier fields. To calculate the matching degree, we employ the stable matching algorithm, generating estimated values for standard compliance evaluation indicators. The performance of our multi-view graph alignment model is assessed on the publicly available dataset DBP15K, and the results demonstrate the superiority of our proposed method over the original approach. Consequently, we apply our method to a customized enterprise dataset to validate the effectiveness of mismatch discovery and compliance evaluation.