Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 10 views

    5 - Cia 2 Key

    Uploaded by

    Ashish Chowdarty
    The document describes computing PageRank and performing similarity calculations on user-item matrices. It provides examples of: 1) Constructing a transition matrix for a small web graph and computing the PageRank vector through iterative multiplication. 2) Calculating Jaccard, cosine, and normalized cosine distances between users based on a utility matrix of item ratings. 3) Recoding ratings in the utility matrix as binary values and recomputing similarity metrics.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    5 - Cia 2 Key

    Uploaded by

    Ashish Chowdarty
    10 views4 pages
    The document describes computing PageRank and performing similarity calculations on user-item matrices. It provides examples of: 1) Constructing a transition matrix for a small web graph and computing the PageRank vector through iterative multiplication. 2) Calculating Jaccard, cosine, and normalized cosine distances between users based on a utility matrix of item ratings. 3) Recoding ratings in the utility matrix as binary values and recomputing similarity metrics.

    Original Title

    5 - CIA 2 KEY

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document describes computing PageRank and performing similarity calculations on user-item matrices. It provides examples of: 1) Constructing a transition matrix for a small web graph and computing the PageRank vector through iterative multiplication. 2) Calculating Jaccard, cosine, and normalized cosine distances between users based on a utility matrix of item ratings. 3) Recoding ratings in the utility matrix as binary values and recomputing similarity metrics.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    10 views4 pages

    5 - Cia 2 Key

    Uploaded by

    Ashish Chowdarty
    The document describes computing PageRank and performing similarity calculations on user-item matrices. It provides examples of: 1) Constructing a transition matrix for a small web graph and computing the PageRank vector through iterative multiplication. 2) Calculating Jaccard, cosine, and normalized cosine distances between users based on a utility matrix of item ratings. 3) Recoding ratings in the utility matrix as binary values and recomputing similarity metrics.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 4

    Key : INT404 Big Data Analytics CIA II Oct-2021

    1.Compute PageRank vector of the web graph for the following:


    Suppose you have a small internet consisting of four web sites www.page1.com,
    www.page2.com, www.page3.com and www.page4.com. Page1.com links with page2.com,
    page3.com and page4.com. Page2.com links with page3.com and page4.com. page3.com
    links with page1.com. Page4.com links with page3.com and page1.com.
    Construct a directed web graph with four nodes, one for each web site. Find a transition
    matrix for the web graph and compute the PageRank vector for the above.

    Let www.page1.com be 1
    Let www.page2.com be 2
    Let www.page3.com be 3
    Let www.page4.com be 4
    0 0 1 1 2
    1 3 0 0 0 

    1 3 1 2 0 1 2
     
    The transition matrix M is 1 3 1 2 0 0
    Initially the importance is uniformly distributed among the 4 nodes, each getting ¼. Denote by V
    the initial rank vector having all entries equal to ¼.
    The new importance vector V1=M V. We can iterate the process and stop whenever the
    updated importance vector is V2=M(MV) = V3=M(M2V).
    1 4  0.25
    1 4  0.25
    V  
    1 4  0.25
       
    1 4  0.25

     0 0 1 0.5 0.25 0.37 


    0.33 0 0 0  0.25  0.08 
    MV     
    0.33 0.5 0 0.5 0.25  0.33
        
    0.33 0.5 0 0  0.25  0.20 
     0 0 1 0.5 0.37   0.43
    0.33 0 0 0   0.08   0.12 
    M 2V  M ( MV )     
    0.33 0.5 0 0.5  0.33 0.27 
        
    0.33 0.5 0 0   0.20   0.16 

     0.43  0.35
     0.12  0.14 
    M V  M (M V )  M 
    3 2  
    0.27  0.29 
       
     0.16  0.20 
     0.35 0.39 
    0.14   0.11
    M 4V  M ( M 3V )  M   
    0.29  0.29 
       
    0.20  0.19 
    0.39  0.39 
     0.11  0.13
    M V  M (M V )  M 
    5 4  
    0.29  0.29 
       
    0.19  0.19 
    0.39   0.38
     0.13  0.13
    M 6V  M ( M 5V )  M   
    0.29  0.29 
       
    0.19  0.19 
     0.38  0.38
     0.13 0.12 
    M V  M (M V )  M 
    7 6  
    0.29  0.29 
       
    0.19  0.19 
     0.38  0.38
    0.12  0.12 
    M 8V  M ( M 7V )  M   
    0.29  0.29 
       
    0.19  0.19 

    Since the last two vectors values are same, we can stop the process and the page ranks for the
    four web sites are 0.38, 0.12, 0.29 and 0.19.
    2. The given utility matrix M for user-item pair is

    User/item a b c d e f g h
    A 4 5 5 1 3 2
    B 3 4 3 1 2 1
    C 2 1 3 4 5 3
    a)For a boolean utility matrix, compute the Jaccard distance between each pair of users.
    The Jaccord distance between (A,B) = 0.5, (A,C) = 0.5, (B,C) = 0.5

    (b)Compute the cosine distance between each pair of users.


    Cosine of the angle between (A,B)=0.601, (A,C) = 0.615, (B,C) = 0.514

    (c)Normalize the matrix by subtracting from each nonblank entry the average value for its user.
    Compute the cosine distance between each pair of users.
    User/item a b c d e f g h Average
    A 4 5 5 1 3 2 20/6
    B 3 4 3 1 2 1 14/6
    C 2 1 3 4 5 3 18/6

    The utility matrix after normalizing ratings:

    User/item a b c d e f g h Average
    A 4/6 10/6 10/6 -14/6 -2/6 -8/6 20/6
    B 4/6 10/6 4/6 -8/6 -2/6 -8/6 14/6
    C -6/6 -12/6 0 6/6 12/6 0 18/6
    Calculations for the cosine of the angle between AB, AC, and BC.

    AB= 0.584, AC= -VE, BC= -VE

    (d)Recode the ratings (3,4,5) as 1 and (1,2 and blank) as 0. Compute the cosine distance
    between each pair of users.
    User/item a b c d e f g h Average
    A 1 1 0 1 0 0 1 0 20/6
    B 0 1 1 1 0 0 0 0 14/6
    C 0 0 0 1 0 1 1 1 18/6
    The Jaccord distance between

    (A,B) = 1- 2/5 = 3/5 = 0.6

    (A,C) = 1 – 2/6 = 0.66


    (B,C) = 1- 1/6 = 5/6 = 0.83

    Cosine Similarity among A,B, and C

    Sim(A,B) = 0.5773

    Sim(A,C) = 0.5

    Sim(B,C) = 0.2886

    You might also like