Social Media Mining: An Introduction

(Axel Boer) #1

P1: Trim: 6.125in×9.25in Top: 0.5in Gutter: 0.75in
CUUS2079-09 CUUS2079-Zafarani 978 1 107 01885 3 January 13, 2014 17:28


262 Recommendation in Social Media

Table 9.4.User-Item Matrix

Lion King Aladdin Mulan Anastasia
John 4 3 2 2
Joe 5 2 1 5
Jill 2 5? 0
Jane 1 3 4 3
Jorge 3 1 1 2

Example 9.5.Consider the user-item matrix in Table9.4and the following
adjacency matrix denoting the friendship among these individuals.

A=



⎢⎢


⎢⎢


⎢⎢




John Joe Jill Jane Jorge
John 0100 1
Joe 1010 0
Jill 0101 1
Jane 0010 0
Jorge 1010 0


⎥⎥


⎥⎥


⎥⎥




, (9.57)


We wish to predict rJill,Mulan. We compute the average ratings and simi-
larity between Jill and other individuals using cosine similarity:

r ̄John=

4 + 3 + 2 + 2


4


= 2. 75. (9.58)


r ̄Joe=

5 + 2 + 1 + 5


4


= 3. 25. (9.59)


r ̄Jill=

2 + 5 + 0


3


= 2. 33. (9.60)


r ̄Jane=

1 + 3 + 4 + 3


4


= 2. 75. (9.61)


r ̄Jorge=

3 + 1 + 1 + 2


4


= 1. 75. (9.62)


The similarities are

sim(Jill,John)=

2 × 4 + 5 × 3 + 0 × 2



29



29


= 0. 79. (9.63)


sim(Jill,Joe)=

2 × 5 + 5 × 2 + 0 × 5



29



54


= 0. 50. (9.64)


sim(Jill,Jane)=

2 × 1 + 5 × 3 + 0 × 3



29



19


= 0. 72. (9.65)


sim(Jill,Jorge)=

2 × 3 + 5 × 1 + 0 × 2



29



14


= 0. 54. (9.66)

Free download pdf