Information Technology Reference
In-Depth Information
N
k
[
F
n
(
K
w
[
n
])]
N
[
F
n
(
K
w
[
n
])]
+
1
Repeat
l
3
Repeat
l
2
Loop
l
4
:
i
1 ~
n
If
N
k
[
i
]
>
T
k
K
wf
[
j
]
K
w
[
F
k
(
N
k
[
i
]) ]
j
j
+
1
Repeat
l
4
Loop
l
5
:
i
1 ~
n
Loop
l
6
:
j
1 ~
n
If
C
l
[
i
]
k
wf
[
j
]
≠ ψ
R
l
[
k
]
C
l
[
i
]
k
k
+
1
Repeat
l
6
Repeat
l
5
End
According to this algorithm, we can easily obtain representative land-
marks. The first procedure in this algorithm is to collect strings of candidate
landmarks from the GPS. The procedure will not end until all landmarks
near the visitors have been transferred in a given radius. Two loops are used
for calculating frequencies of keywords in candidate landmarks. By making
intersections of landmarks, the keywords can be easily obtained. With a given
threshold of frequency, unnecessary keywords will be filtered, remaining
ones can be used for finding representative landmarks. The final part of this
algorithm is using the selected keywords to find landmarks that are related
to users. By comparing obtained keywords and candidate landmarks, strings
of landmarks composed of substrings with low frequency will be filtered.
In the algorithm, frequently appearing keywords in landmarks will be
selected. Two kinds of results have different meanings to visitors. The first
kind of keywords indicates where users are. Such keywords help users to
obtain information of this area via an agent. The second kind of keywords
represents objects like stores or restaurants. If there are buildings with the
same purpose located in the area, keywords of these objects will also be
considered as representative landmarks. The two types of landmarks are
obtained by calculating frequencies. It is probably not enough for agents to
gather all the information visitors need, users' comments are also a kind of
information that can be used for discovering objects or landmarks besides
results generated from this algorithm. This is not included in the algorithm
but will be discussed in the next section.
13.2.2 Similar Travel route Search
In the previous section, we introduced an approach to find related informa-
tion by filtering keywords from landmarks near visitors. On a journey, visitors
probably travel to several locations. In the selecting_representative_landmark
approach, only landmarks in the same area are processed. Since visitors may
