Information Technology Reference
In-Depth Information
Algorithm 1.
Subscription Installation
Require:
sub
//the subscription
vector
<
FrequentItemset
>
FIS
[
0
...
n
−
1
]
//the set of frequent itemsets
vector
<
FrequentItemset
>
VFIS
[
0
...
m
−
1
]
//the set of virtual frequent itemsets
ext ract
(
se
)
//returns attribute names in se
IFContain
(
A
,
F
)
//returns TRUE if A contains F, FALSE otherwise
LexiOrder
(
Att r
)
//returns a string by bunching the names of attributes in the attribute set
lexicographically
send
(
message
,
key
)
//deliver message to the node whose ID closet or immediately suc-
ceeds to key
1: set
Att rSet
=
NU LL
;
2: set
support
=
1;
3: set
num
=
0;
4: set
Att r
sub
=
ext ract
(
sub
)
;
do
6:
if
IFContain
(
Att r
sub
,
FIS
[
i
]
.
it emset
)==
true
then
7:
num
++;
8:
if
FIS
[
i
]
.
support
<
support
then
9:
Att rSet
=
FIS
[
i
]
.
it emset
;
10:
support
=
FIS
[
i
]
.
support
;
11:
end if
12:
end if
13:
end for
14:
if
num
==
0
then
15:
i
=
RandomInteger
(
0
,
m
−
1
)
;
16:
Att rSet
=
VFIS
[
i
]
.
it emset
;
17:
end if
18:
key
=
hash
(
LexiOrder
(
Att rSet
))
;
19:
send
(
sub
,
key
)
;
5:
for
each
i
∈
[
0
,
n
−
1
]
N
strings will be generated by bunching the names of attributes in each attribute
set lexicographically. Finally, the event is sent to the
N
related nodes in the overlay
network for matching.
The event publication policy ensures that all subscriptions that match with an
event are found. Moreover, because all subscriptions that don't contain frequent
itemset are stored on no more than
M
RPs, the number of events that are sent for
publication is reduced dramatically.
3.5
Subscriptions and Events Matching
When a RP receives an event
e
, it matches
e
with all the subscriptions stored on it
and finds the matched subscribers. In DMPSS, subscriptions that contain the same
frequent itemset can emerge strong covering relations, so covering-based matching
algorithms are suggested to be used to improve matching efficiency.
Search WWH ::
Custom Search