Information Technology Reference
In-Depth Information
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
2
4
6
8
10
URI Number
Fig. 5.9
Xaxis
depicts the URI used in the experiment,
Yaxis
depicts the different D Statistics
from the KS Test. The
dotted line
is the 'no tag suggestion' condition, while the
solid line
is the
'tag suggestion' condition
suggestions do not only
not
cause the power-law to form, but it seems that they
somehow prevent it from being formed. On the other hand, the 'no tag suggestion'
condition results in a significantly good fit to a power-law. Therefore, the result
is somewhat counter-intuitive, as according to our experimental data a simple tag-
based suggestion mechanism is unlikely the main cause of the power-law formation.
5.3.4
Influence of Tag Suggestions on the Tag Distribution
Given that the KS test shows that there is a significant and perhaps counter-intuitive
difference in the emergence of the power-law distributions between the conditions,
we need a more fine-grained way to tell what the differences are in the distributions
for the two conditions.
5.3.4.1
Ranked frequency distribution
In order to observe the micro-behavior of the 'tag suggestion' and 'no tag sug-
gestion' distributions, we investigate whether or not the tag suggestion tags are
'forced' higher in the distribution, so leading to a more sparse long tail and an
Search WWH ::
Custom Search