490
Q. Xu et al.
(2) The structure of entity relationship is difficult to change dynamically;
(3) The depth-first search traversal ability is insufficient to carry out association
reasoning.
3 ATM Knowledge Graph-Based Fusion and Correlation
Analysis Method
In order to get all related feature objects on breadth and get relative path on
depth, an ATM knowledge graph-based fusion method is proposed to reorganize the information, and after that, a breadth-first and depth-first search-based
correlation analysis method is designed to achieve the required goal.
3.1 ATM Knowledge Graph-Based Fusion Method
In this work, we first build the ATM knowledge graph to reorganize all the
important ATM features, such as Airport, Route, Airspace, Flight Plan, Airline,
Land Mark, Control Unit, and so on. The graph data structure is chosen to
represent and save the relationship between all the ATM feature instances, which
is different from the E-R divided tables storage method.
The ATM knowledge graph G consists of a collection of nodes N and edges
E, which is G = (N, E). The nodes represent all the ATM feature instances and
the edges represent the relationship between different feature instances. Figure 2
gives a partial fragment of how the ATM knowledge graph looks like.
In the graph, all the ATM features are organized together and all the relationships between each feature instance are saved as edges. With the graph, we
Fig. 2. A partial fragment of the ATM knowledge graph. Different colors of the nodes
represent different feature type and the edge between each two nodes has its semantic
meaning.