An ATM Knowledge Graph-Based Method of Dynamic . . .
493
Table 1. The breadth-first search results of different feature type.
Searched
feature
instance
Feature
type
Distance from searched feature instance
1 (direct connect)
2
3
Feature
numbers
Elapsed
time (ms)
Feature
numbers
Elapsed
time (ms)
Feature
numbers
Elapsed
time (ms)
Shanghai
Pudong
Airport 146
58
905
113
3873
258
CSN3173 Flight
8
33
1020
126
2959
283
A593
Route
27
4
55
6
296
36
DOGAR Route
point
84
8
413
39
3150
198
Table 2. The depth-first search results of relative path.
First feature
instance
Second feature
instance
Relative
path
numbers
Shortest
length
Longest
length
Search elapsed
time (ms)
Shanghai
Pudong airport
CSN3173
85
3
3
702
CCA4228
CES2471
15
2
6
453
Shanghai
Pudong airport
Beijing Capital
airport
124
2
8
1560
Table 1 shows the case of BFS results of different feature type. We test the
distance (1, 2, 3) from searched feature instance to verify the efficiency of the BFS
method. Distance 1 represents those feature nodes connect with searched feature
node directly, distance 2 represents those nodes connect through two edges to the
searched feature node, and so on. The result shows that the proposed method
can find the same amount of direct-connected feature nodes more efficiently and
flexibly, besides can find more deeper relative feature nodes which is hard to
achieve by the traditional method.
Figure 4 shows an example of a breadth-first search result of route A593 with
the distance (1, 2, 3).
Table 2 shows the case of depth-first search results of relative path. We choose
any two feature instances to get their relative paths which cannot be achieved
by the traditional method. The result shows that there may be more than one
relative path between two feature instances, some of the paths are useful but
some others are not. Further research work is needed to describe the path from
semantic view.
Figure 5 shows the example of a depth-first search result of relative path
between Shanghai Pudong international airport and flight CSN3173. We can
493
Table 1. The breadth-first search results of different feature type.
Searched
feature
instance
Feature
type
Distance from searched feature instance
1 (direct connect)
2
3
Feature
numbers
Elapsed
time (ms)
Feature
numbers
Elapsed
time (ms)
Feature
numbers
Elapsed
time (ms)
Shanghai
Pudong
Airport 146
58
905
113
3873
258
CSN3173 Flight
8
33
1020
126
2959
283
A593
Route
27
4
55
6
296
36
DOGAR Route
point
84
8
413
39
3150
198
Table 2. The depth-first search results of relative path.
First feature
instance
Second feature
instance
Relative
path
numbers
Shortest
length
Longest
length
Search elapsed
time (ms)
Shanghai
Pudong airport
CSN3173
85
3
3
702
CCA4228
CES2471
15
2
6
453
Shanghai
Pudong airport
Beijing Capital
airport
124
2
8
1560
Table 1 shows the case of BFS results of different feature type. We test the
distance (1, 2, 3) from searched feature instance to verify the efficiency of the BFS
method. Distance 1 represents those feature nodes connect with searched feature
node directly, distance 2 represents those nodes connect through two edges to the
searched feature node, and so on. The result shows that the proposed method
can find the same amount of direct-connected feature nodes more efficiently and
flexibly, besides can find more deeper relative feature nodes which is hard to
achieve by the traditional method.
Figure 4 shows an example of a breadth-first search result of route A593 with
the distance (1, 2, 3).
Table 2 shows the case of depth-first search results of relative path. We choose
any two feature instances to get their relative paths which cannot be achieved
by the traditional method. The result shows that there may be more than one
relative path between two feature instances, some of the paths are useful but
some others are not. Further research work is needed to describe the path from
semantic view.
Figure 5 shows the example of a depth-first search result of relative path
between Shanghai Pudong international airport and flight CSN3173. We can