A construction method of a multidimensional and multilingual association network for earth surface system science data

The use of association networks to calculate metadata similarity for data association and recommendation is a well-established approach. However, challenges persist in the field of open scientific data association within the Earth surface system. Existing methodologies are primarily designed for single data sources, limited dimensions, and specific linguistic domains, limiting their applicability and effectiveness. Enhancing the efficiency and adaptability of association algorithms is crucial for further advancements in this field. In this paper, we propose a multidimensional and multilingual association method for Earth surface system scientific data to support precise data retrieval and intelligent recommendation. The method quantitatively computes one-dimensional and multidimensional similarities using five primary indicators of multilingual data, enabling semantic interconnection and facilitating knowledge discovery within open scientific data related to the Earth surface system. We obtained metadata from several authoritative sources, including NESSDC, NASA, and Data.gov, and performed association calculations, constructed multidimensional association networks, and evaluated them for optimization. The key strengths of this study include its ability to construct one-, two-, and multidimensional association networks effectively, support cross-lingual and cross-standard data interconnections, and assess network node importance. Finally, we conducted a comparative analysis with classical methods, demonstrating that our approach achieves higher accuracy and better scalability.