Identification of Duplicate Features Among Universal Hydrographic Data Models to Enhance Interoperability Through Natural Language Processing

In the field of geographic information, research on identifying and integrating duplicate data for quality improvement has been actively conducted, contributing to enhancing the reliability of spatial information-based decision-making. The International Hydrographic Organization (IHO) has standardized hydrographic data through a universal hydrographic data model in the hydrographic field. As multiple institutions are independently modeling the data, duplication is occurring in the process of defining ‘features’, which abstract real-world objects as classes. Quality degradation due to these duplicate features can cause confusion in the decision-making process of smart ships, potentially threatening safe navigation. Previous studies in the hydrographic field have identified duplicate features manually, but continuous updates and an increasing number of products have made it difficult to identify all duplicate features. This study proposes a method to identify duplicate features defined in data models using deep learning-based natural language processing models. Through the proposed method, 313 additional duplicate features that were not previously discovered were identified. This efficient approach to identifying duplicate features is anticipated to support interoperability processes, thereby contributing to the stable operation of smart ships and the enhancement of maritime safety.