Finding Frequent Webpage Access Patters

1
Finding Frequent Webpage Access Patters

• Part 2 Searching for Frequent Trees
• Jennifer Cuzzort
• Ashley Plier

2
URLs List Represented as a Tree

• If a User browses the internet without using the
  back button, the URLs visited can be represented
  as a sequence.
• If a User browses the internet using the back
  button, the accessed URLs can be represented as a
  tree.

3
Tree Representation and Terminology
4
Tree Written Representation

• The tree on the previous slide can be represented
  in the following format
• A(B(EFG)C(HI)D(JK))
• Notice that each nodes children are listed
  inside a set of parentheses following their
  parent node
• The first generation node is listed first,
  outside of any parentheses

5
Method of Implementation

• The information from the file that is to be
  searched is stored in a 2-D character array
• The program searches for the location of each
  occurrence of a character, beginning with A,
  inside the array
• The location the balance parentheses following
  each occurrence of A is stored in a vector
  called FoundData

6
Method of Implementation

• When all the occurrences of A are found and
  stored in FoundData, the program stores the
  location of all the characters that appear inside
  the balanced parentheses of the records in
  FoundData
• This information is stored in FoundData2

7
Example Searching for B

• FoundData2
• C (6, 6) on line 1
• D (6, 6) on line 1
• A (4, 7) on line 2
• D (12, 12) on line 2
• K (6, 6) on line 2
• L (11, 11) on line 2
• M (7, 7) on line 2
• P (9, 12) on line 2

• FoundData
• B (3, 6) on line 1
• B (10, 10) on line 1
• B (3, 13) on line 2

8
Method of Implementation

• The program uses the content of FoundData and
  FoundData2 to build all the trees with a first
  generation node B in the array
• The information is stored in a vector called
  Trees

9
Example Searching For B

• Trees
• B(CD) on line 1
• B(C) on line 1
• B(D) on line 1
• B(A(KM)) on line 2
• B(R(LD) on line 2
• B(A) on line 2
• B(A(K)) on line 2
• B(A(M)) on line 2
• B(K) on line 2

• B(M) on line 2
• B(KM) on line 2
• B(R) on line 2
• B(R(L)) on line 2
• B(R(D)) on line 2
• B(L) on line 2
• B(D) on line 2
• B(LD) on line 2

10
Method Of Implementation

• The occurrence of each tree in the vector is
  counted
• If the tree appears on more than the minimum
  support number of lines, then it is considered
  frequent
• Frequent trees are added to a vector called
  FinalTrees

11
Method of Implementation

• The program repeats the previous steps untill it
  has checked for all characters A through Z
• At this point FinalTrees contains all of the
  frequent trees that occur in the input file
• The content of trees needs to be filtered for a
  more concise output file

12
Method of Implementation

• The contents of FinalTrees is printed to an
  intermediate output file called temp.txt
• Then the information in temp.txt is stored in a
  second 2-D array
• The trees that do not appear as part of a larger
  tree are recorded in a new vector
• Trees that do appear in a larger tree are ignored

13
Example Filtering the Trees in Temp.txt

• Temp.txt
• A(B(C(DE)))
• A(C(E))
• A(B(F))
• B(D)
• E(SJ)
• E(S)
• E(F(G(H)))
• F(H)

• Output.txt
• A(B(C(DE)))
• A(B(F))
• E(SJ)
• E(F(G(H)))

14
Conclusion

• Using the method described the Tree searching
  part of the Data Mining program can effectively
  evaluate the frequent trees in a list of URLs
• The results of this evaluation are concisely
  reported in the output file