The probability of transition is calculated by theratio of the number of times the correspondingsequence of pages was traversed and the number of times a hyperlink page was visited. A state of a page is composed by two other states Startstate(S) and Terminal State (F).Probability of hyperlink is based on the contentof page being viewed. Navigation matrix is asfollows:This Indicate navigation control can reach attotal 10 times at T.
A B C D E F T
A
0
3 / 51 / 2
0 0 0 0
B
0 0
1 / 2
0 0 0 0
C
0 0 0
1 1 / 22 / 5
0
D
0 0 0 0
1 / 2
0
1 / 5
E
0 0 0 0 0
3 / 53 / 10
F
0 0 0 0 0 0
1 / 2
T
0 0 0 0 0 0
1
Table 2: frequency of each Node and theirprobability.
So we can identify that total probability of visitof A is 8/39, B is 5/39, C is 10/39, D is 5/39, Eis 6/39 and F is 5/39.Here
NP
i j
is a navigation probability matrix where NP is the probabilitywhere next stage will be
j
. Navigation probability is defined as NP
i j
0,1
And for all j NP
i j
=1. The initial probability of astate is estimated as the how many number of times a page was requested by user so we cansay that every state has a positive probability.The Traditional Markov model has somelimitations which are as follows.1.
Low order Markov Models has goodcoverage but less accurate due to poor history.2.
High order Markov Models suffersfrom high state space complexity.In higher-order Markov model number of statesexponential increases as increase in the order of model. The exponential increment in number of states increases search space and complexity Higher-order Markov model also have low coverage problem. In proposed model, each request with itstime-duration is considered as a state. A session is asequence of such states. The m-step Markov modelassumes that the next request depends only on last mrequests. Hence, the probability of the next request iscalculated byP(r
n+1
|r
n...r1
) = P(r
n+1
|r
n...
r
n
−
m +1
),Where r
i
is the i th request in a session, i=1, 2... n, r
n
is the current request, and r
n+1
is the next request.From this equation, if m=1 (the 1-step model), thenext request is determined only by the currentrequest [5]. The Matrix CM is of conditional probability of previous occurrence. The state matrixCM is a square matrix. So we need to be calculatingthe probability of each page. So we need to design amodel that is dynamic in nature means prediction is based on the next incoming and outgoing node. TheMarkov model construction starts with the first rowof table (first navigation pattern)
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 9, No. 7, July 2011126http://sites.google.com/site/ijcsis/ISSN 1947-5500