This is what I got:

(a)

S1: <p1,p2,p5,p3> <p1,p2,p7,p3>

S2: <p1,p4,p5,p1,p6,p7>

S3: <p1,p6,p1,p4>

S4: <p5,p1,p4> <p5,p1,p6,p7>

S5: <p5,p3>

S6: <p1,p2,p7,p3>

S7: <p2,p7>

S8: <p1,p2,p4,p1,p2,p6,p7,p3>

 So the session is:

<p1,p2,p5,p3>

<p1,p2,p7,p3>

<p1,p4,p5,p1,p6,p7>

<p1,p6,p1,p4>

<p5,p1,p4>

<p1,p2,p4,p1,p2,p6,p7,p3>

<p1,p2>

<p1,p4>

HelloWorld wrote:

This algorithm is written very confusing..

6: ForEach Page(i) in CandSession

where is the EndFor? I tried to use Ruler to find it, and I couldn't find it..

There are three steps in this algorithm.

Step I starts on line 5: TPageSet := {}

Then For-loop runs till line 13th inclusive.

In this for-loop the TPageSet is being updated.

Then, for-loop shall be closed and then (on line 15) Step II begins and the CandSession is updated.

Just looking on this code you can see that this for-loop cannot run any further because a few lines later another for-loop (also by Page_i) begins.

Step III-a will start if condition on line 16 is sutisfied.

Otherwise, Step III-b will start (line 19).

HelloWorld wrote:

 

Then For-loop runs till line 13th inclusive.

the problem is that there's no EndFor for the for loop that starts on Line 6.. Are you sure though?

I understand there's not EndFor line.

Here is the code. CandSession is a scope for that loop, and CandSession is changing in line 15. Therefore, the loop must end before that line. On the other hand, the last use of Page_i by this loop is on line 13. That means EndFor line shall be between 13 and 15. I added it on line 14 (comments don't need to be numbered).

 6: ForEach Pagei in CandSession
 7:   StartPageFlag := TRUE
 8:   ForEach Pagej in CandSession with j > i
 9:     If ( Link[Pagei, Pagej] = true) and ( TimeDiff(Pagej , Pagei) ≤ σ ) Then
10:       StartPageFlag := FALSE
11:   End For
12:   If StartPageFlag = TRUE Then
13:      TPageSet := TPageSet U {Pagei}
14: End For
// Remove the selected pages from the current seq.
15: CandSession := CandSession − TPageSet 

There is another possible mistake in line 9:

9: If ( Link[Page_i, Page_j] = true) ...

Here, Link[Page_i, Page_j] means that there are link directed from i to j. However, it would make a sense if we switch i and j.

HelloWorld wrote:

Here, Link[Page_i, Page_j] means that there are link directed from i to j. However, it would make a sense if we switch i and j. 

yeah, i agree with you.. but i still don't see on why if there's no link between j to i, would make page i as a start page.. it's basically saying that subpages cannot have link back to the main page.. doesn't make sense..

Actually, it does not say that subpages cannot have link back to the main page (back to the page we just came from). It just chooses not to use such links. As I understand, we are interested in forwarding links to move away from Page_i, along the session. So it is not important for us to know that there is a way to return back to the main page through the same way. I just don't see another explanation for that.

TSessionSet := TSessionSet ∪ {[Page_i]}

TSessionSet stands for the Temporary Session Set. In Step III we are appending TPageSet to TSessionSet, but because at this moment (Step III-a) TSessionSet is an empty set, we can simply do Union.

We donot union to anything.  We union it to an empty set.  And this only occurs during the first step

HelloWorld wrote:

TSessionSet stands for the Temporary Session Set. In Step III we are appending TPageSet to TSessionSet, but because at this moment (Step III-a) TSessionSet is an empty set, we can simply do Union.

I know that it's union and that it's Temporary Sesion Set, my question is more, it's being unioned to what? I thought it's being unioned to the session that has that Pagei, but i'm not 100% sure..

This is what I am getting:

Sessions:

a)

{P1, P2, P5,P3}

{P1,P2,P7,P3}

{P1, P4,P5,P1,P6,P7}

{P1,P6,P1,P4}

{P5,P1,P6,P7}

{P4,P1,P6,P7}

{P1,P2,P4,P1,P2,P6,P7,P3}

b) frequent patterns:

P1,P2

P1,P6

P1,P6,P7

I think we are getting so different results because we read the algorithm code in different ways.

It was my mistake to add the EndFor between IF's. The more appropriate way is the following:

 6: ForEach Pagei in CandSession
 7:   StartPageFlag := TRUE
 8:   ForEach Pagej in CandSession with j > i
 9:     If ( Link[Pagei, Pagej] = true ) and ( TimeDiff(Pagej , Pagei) ≤ σ ) Then
10:       StartPageFlag := FALSE
11:     If ( StartPageFlag = true ) Then
12:       TPageSet := TPageSet U {Pagei}
13:   End For
14: End For
// Remove the selected pages from the current seq.
15: CandSession := CandSession − TPageSet 

But <p1,p4,p1,p6,p7> is a subsequence of <P1,P4,P5,P1,P6,P7>, and Maximality Rule will be violated.

cysun wrote:

No one is completely correct so far. From S2 you should get <P1,P4,P5,P1,P6,P7> and <P1,P4,P1,P6,P7>.

Because in your slides (more.ppt), page 5

A=<a1a2a3…an>

B=<b1b2b3…bm>

A is a subsequence of B if there exists 1<=j1<j2<…<jn <=m such that a1 'belong' bj1,a2 'belong' bj2,…,an 'belong' bjn

And it doesnot say that all elements should be adjacent.

cysun wrote:

xieguahu wrote:

But <p1,p4,p1,p6,p7> is a subsequence of <P1,P4,P5,P1,P6,P7>, and Maximality Rule will be violated.

cysun wrote:

No one is completely correct so far. From S2 you should get <P1,P4,P5,P1,P6,P7> and <P1,P4,P1,P6,P7>.

 

Why <p1,p4,p1,p6,p7> is a subseqence of <p1,p4,p5,p1,p6,p7>? I haven't noticed a clear definition for sequence containment in the paper, so I assume that to be a subsequence, all elements must be adjacent.

cysun wrote:

xieguahu wrote:

Because in your slides (more.ppt), page 5

A=<a1a2a3…an>

B=<b1b2b3…bm>

A is a subsequence of B if there exists 1<=j1<j2<…<jn <=m such that a1 'belong' bj1,a2 'belong' bj2,…,an 'belong' bjn

And it doesnot say that all elements should be adjacent.

cysun wrote:

xieguahu wrote:

But <p1,p4,p1,p6,p7> is a subsequence of <P1,P4,P5,P1,P6,P7>, and Maximality Rule will be violated.

cysun wrote:

No one is completely correct so far. From S2 you should get <P1,P4,P5,P1,P6,P7> and <P1,P4,P1,P6,P7>.

 

Why <p1,p4,p1,p6,p7> is a subseqence of <p1,p4,p5,p1,p6,p7>? I haven't noticed a clear definition for sequence containment in the paper, so I assume that to be a subsequence, all elements must be adjacent.

That definition does not apply to this paper.

But you are right that from S2 there's only one session <p1,p4,p5,p1,p6,p7>. I tried the algorithm again and realized that I made a mistake. Sorry about that.

I think it is not correct, should it be <p5,p1,p6,p7><p4,p1,p6,p7>?

cysun wrote:

xieguahu wrote:

...

S4: <p5,p1,p4> <p5,p1,p6,p7>

...

How did you get <p5,p1,p4> for S4? There's no p1 proceeding p4 in S4. The sessions for S1, S2, and S3 are correct. I'm continuing to verify the rest.

xieguahu wrote:

I think it is not correct, should it be <p5,p1,p6,p7><p4,p1,p6,p7>?

It should be just <p5,p1,p6,p7>. p4 is not linked from p5, and according to the algorithm, it's not included in the new sessions.

alomo wrote:

cysun wrote:

xieguahu wrote:

I think it is not correct, should it be <p5,p1,p6,p7><p4,p1,p6,p7>?

It should be just <p5,p1,p6,p7>. p4 is not linked from p5, and according to the algorithm, it's not included in the new sessions.

I think xieguahu is right. The algorithm says that in case the Link(P_n, P_n+1) = false we have to check pages in the constructed part of this session that visited before P_n (going backward from P_n-1 to P₁) to find a link to P_n+1. An if there is no link found, the page P_n+1 becomes the first page of a new session. Therefore, we will have two sessions that xieguahu found - <P₅,P₁,P₆,P₇> and <P₄,P₁,P₆,P₇>.

Yes, you are right.

alomo wrote:

...

<P₁, P₂, P₄> + <P₃> -> <P₁, P₂, P₄>, <P₃> (here a new session started because non of pages P₁, P₂, or P₄ linked to the P₃;

...

this is incorrect - <P₃> doesn't become a new session. You can read line 20-30 of Algorithm 1 to see.