Design and Implementation of Satellite Knowledge Graph and Its Application

Knowledge graphs (KGs), representing multi-relational data as a semantic graph from structured information stored in triples, have attracted wide attention in industrial and academic communities. The previous study mainly focuses on the construction techniques of KG from semantically structured information; it is still challenging to understand how these KGs are being used for different applications and what methods they employ. Thus, we provide both design and implementation techniques of KG and its applications in relational query and question-answering systems. The applications implemented on satellite KG consists of over 5000 nodes and 38,000 relationships extracted from the union of concerned scientists (UCS) satellite database, gunter's space page, and other websites. For the satellite question-answering system, we used the NLTK tool with the NLP technology to segment and analyze the given questions that were trained by using the convolutional neural network model over the constructed satellite KG. Our proposed method presents a satellite KG and its application in a form that can be easily handled for further research and analysis.