Ontology-based query graph matching

Abstract

The requirement for business competition among organizations has become more complicated in the past decade. One of these issues is how to react and response to queries from clients with the aim of producing quick and timely decision-making. To address this problem, this paper proposes a framework for identifying the frequent queries from multiple sources. The main contribution of this work is (1) query matching based on the ontology. In fact, query matching has been widely explored using various methods, most of which are based on a labeling approach and utilize isomorphism to map a query to subgraphs in a graph database. However, these traditional methods may not capture enough semantic similarity when mapping terms together. Ontology-based matching can address this issue more effectively than mere labeling isomorphism. In this work, three pruning techniques and their combination within dynamically weighted sets of vertex-associated elements are proposed. In addition, (2) a subgraph/subontologies clustering technique is recommended to extract the results of the framework. This technique is based on an existing clustering method, but requires improvements in the data processing step to be applicable to clustering subontologies/subgraph objects. Finally, (3) we experimentally verify the effectiveness and efficiency of our pruning techniques using synthetic data and compare the techniques themselves.