A Deep Knowledge-Based Evaluation of Enterprise Applications Interoperability

Abstract

Enterprise is a dynamic and self-managed system, and the applications are an integral part of this complex system. The integration and interoperability of enterprise software are two essential aspects that are at the core of system efficiency. This research focuses on the interoperability evaluation methods for the sole purpose of evaluating multiple enterprise applications interoperability capabilities in the model-driven software development environment. The peculiarity of the method is that it links the causality modeling of the real world (domain) with the traditional MDA. The discovered domain causal knowledge transferring to CIM layer of MDA form the basis for designing application software that is integrated and interoperable. The causal (deep) knowledge of the subject domain is used to evaluate the capability of interoperability between software components. The management transaction concept reveals causal dependencies and the goal-driven in-formation transformations of the enterprise management activities (an in-depth knowledge). An assumption is that autonomic interoperability is achievable by gathering knowledge from different sources in an organization, particularly enterprise architecture, and software architecture analysis through web services can help gather required knowledge for automated solutions. In this interoperability capability evaluation research, 13 different enterprise applications were surveyed. Initially, the interoperability capability evaluation was performed using four know edit distance calculations: Levenshtein, Jaro-Winkler, Longest common subsequence, and Jaccard. These research results are a good indicator of software interoperability capability. Combining these results with a bag of words library gathered from “Schema.org” and included as an addition to the evaluation system, we improve our method by moving more closely to semantic similarity analysis. The prototype version for testing of enterprise applications integration solution is under development, but it already allows us to collect data and help research this domain. This research paper summarizes the conclusions of our research towards the autonomic evaluation of interoperability capability between different enterprise applications. It reveals basic concepts on which we proved our assumption that enterprise application could be evaluated in a more objective, calculable manner.