Information Technology Reference
In-Depth Information
A landmark selection engine is also designed for such a purpose. A
detailed description about implementation of this system will be discussed
in the next section. On a journey, visitors always take photos of the same
region. By functions provided by a GPS on the users' mobile devices, we
can obtain some landmarks. These landmarks may help users to find related
information through the travel topic agent. Not all landmarks around visi-
tors can be referenced for gathering information that users are interested in.
For an example, visitors arrive in a region with restaurants, souvenir stores,
and parking lots. They may be interested in information other than parking
lots. There may be common landmarks in the same region. In order to get
an adequate string of landmarks where visitors are located, we developed
an algorithm to find keywords filtered from all landmarks in the region for
further search via the travel topic agent. The algorithm and variable deini-
tions are
•
l
: A list composed of candidate landmarks. All landmarks around
visitors will be assigned to this list.
R
C
•
l
: A list composed of representative landmarks. By procedures of
this algorithm, several landmarks will be filtered and stored in this
list.
G(R
• )
:
With a given radius R
,
the GPS libraries will return a string of
landmarks nearby.
K
•
w
: A list composed of keywords from the intersection of candidate
landmarks
C
l
.
N
•
k
: A list composed of frequent keywords. In this algorithm, the
frequency is assigned as an integer.
F
•
n
(
K
w
): A subfunction to indicate index of
N
k
.
T
•
k
: An integer threshold to distinguish if a keyword is necessary for
filtering representative landmarks or not.
F
•
k
(
N
k
): If a keyword's frequency is larger than the threshold
T
k
, the
function
F
k
(
N
k
) is used to indicate the keyword in the candidate
landmark list Cl.
l
.
K
•
wf
: A list composed of the final results of keywords, by computing
N
k
and
T
k
.
Selecting_representative_landmarks()
Loop
l
1
: i
1 ~
n
C
l
[
i
]
G
(
R
)
Repeat
l
1
until
G
(
R
)
= ψ
Loop
l
2
:
i
1 ~
n
Loop
l
3
:
j
1 ~
n
If
i
≠
j
and If
C
l
[
i
]
∩
C
l
[
j
]
≠ ψ
K
w
[
n
]
C
l
[
i
]
∩
C
l
[
j
]
