Analytics in a Big Data World by Bart Baesens

Author:Bart Baesens [Baesens, Bart]
Language: eng
Format: epub, pdf
Published: 2015-11-04T14:11:03+00:00

S O C I A L N E T W O R K A N A L Y T I C S

◂

123

mining. A popular technique here is the Girvan‐Newman algorithm,

which works as follows: 1

1. The betweenness of all existing edges in the network is calcu-

lated fi rst.

2. The edge with the highest betweenness is removed.

3. The betweenness of all edges affected by the removal is

recalculated.

4. Steps 2 and 3 are repeated until no edges remain.

The result of this procedure is essentially a dendrogram, which can

then be used to decide on the optimal number of communities.

SOCIAL NETWORK LEARNING

In social network learning, the goal is within‐network classifi cation to

compute the marginal class membership probability of a particular node

given the other nodes in the network. Various important challenges arise

when learning in social networks. A fi rst key challenge is that the data are not independent and identically distributed (IID), an assumption often

made in classical statistical models (e.g., linear and logistic regression).

The correlational behavior between nodes implies that the class mem-

bership of one node might infl uence the class membership of a related

node. Next, it is not easy to come up with a separation into a training

set for model development and a test set for model validation, since the

whole network is interconnected and cannot just be cut into two parts.

Also, there is a strong need for collective inferencing procedures because

inferences about nodes can mutually infl uence one another. Moreover,

many networks are huge in scale (e.g., a call graph from a telco pro-

vider), and effi cient computational procedures need to be developed to

do the learning.2 Finally, one should not forget the traditional way of

doing analytics using only node‐specifi c information because this can

still prove to be very valuable information for prediction as well.

Given the above remarks, a social network learner will usually

consist of the following components: 3

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.