Sensor Ontology Matching With Compact Firework Algorithm

Sensor ontology provides a standardized and conceptual framework for the Internet of Things (IoTs), but their practical applications are often hampered by the ontology heterogeneity issue, emerging from varying development contexts and semantic discrepancies. Sensor ontology matching (SOM) addresses this issue by mapping semantically identical entities within sensor ontologies, whose foundation is the utilization of similarity measure (SM). Although SM can quantify the similarities of entities, the complex nature of entity heterogeneity precludes any single SM from consistently excelling across diverse scenarios, therefore combining different SMs becomes essential to enhance the accuracy of matching results. The swarm intelligence (SI) technique, particularly the firework algorithm (FA), shows promising results in SM combination, but its application in matching sensor ontologies is hampered by efficiency issues. To overcome this limitation, this article proposes an innovative compact FA (cFA) for SOM. First, we present an iterative framework for SM combination to optimize the quality of SOM results. Second, we develop a compact encoding mechanism that can increase the overall efficiency of the SOM. Finally, we design a new fitness function that is able to autonomously gauge individual quality and obviate expert intervention. The experiments use the Benchmark and Conference datasets for the ontology alignment evaluation initiative (OAEI) and ten real SOM tasks to evaluate the performance of the designed approach. Comparisons with advanced matching techniques show the effectiveness and the efficiency of the developed method, which consistently achieves high-quality matching results with an average f-measure of 0.80.