what-when-how
In Depth Tutorials and Information
For building the basic network G b and the collaborative learning network G e ,
the following procedures were followed:
Recover all the names of the participants from every message, and for each
topic put them in the same line. he “HAS Centre Browser” is a Windows-
based program that will routinely help us to accomplish this task.
All the members' names are sorted, put in an Excel file, and tagged with
unique sequence numbers.
All the names of the participants are now changed into the unique num-
ber labels, and this is done by programming. he idea here is to put the
names in an appropriate input format that can be used in any social net-
work programs.
Using the UCINET programs, diverse network statistics can be pulled out,
and network graphs can be drawn.
An experiment was conducted from November 1, 2004, to November 14, 2004,
to better understand the online discussion forum. Indeed, with the HAS Centre
Browser, they pulled out messages from 36 forums, and we found 24,384 topics
with 51,724 replies and 53,070 counts of members. he highest number of replies in
a speciic topic was 128. he idea here is to provide some formulas and graphs that
best describe the distribution of replies and the distribution of views in all forums.
he following graph (Figure  16.4) shows the relationship between the fre-
quency distribution and the number of replies for individual topics submitted in
all the 36 forums.
he Y-axis for this graph is chosen to be logarithmic because the frequency
numbers are a combination of both higher and lower numbers; moreover, selecting
a logarithmic Y-axis will allow a better interpretation of their results. Also, they
used an exponential trendline, and it is represented by a straight line in our figure.
Note from this graph that most data fall along a straight line; hence, they folllow
the exponential distribution, whose probability density function P ( k ). = λ∗ e ( −λ∗ k ) ,
where λ = 0.266 represents the best-fitted straight line we can get from the figure.
his implies that in an online discussion forum, the probability of having ignored
(unreplied) messages is very low, and the average messages answered is close to
1/ λ = 3.76.
Now, we investigate the relationship between the number of topic viewers and
the messages submitted in a Web-based discussion environment. his relationship
will help us understand conversations going on inside the forum. Figure 16.5 shows
the graph of the distribution of views in all forums.
After some studies, researchers discovered that the data followed an unsual kind
of network distribution called the lognormal distribution. It can be represented by
1
n k m
S
(
)
2
1
P k
( )
=
) *
e
2
2
k s
* *
( *
2
π
Search WWH ::




Custom Search