Database Reference
In-Depth Information
Fig. 14.7 Distribution of Q ( T ) for featured and nonfeatured articles
where q ( t i ) is the quality of the revision submitted at time t i and n is the total number
of revisions up to time T . Figure 14.7 shows the distribution of Q ( T ) for both feature
and nonfeatured articles. While the average of Q ( T ) is relatively high for both
featured and nonfeatured articles, it is higher for featured articles -74% vs. 65%.
To estimate the proportion of time during which an article is in a high-quality
state, we also define the duration QD( T )by
P 1 ð
t i þ 1
t i Þ
q
ð
t i Þ
QD
ð
T
Þ¼
(14.6)
T
t 1
The distribution of QD( T ) for both featured and nonfeatured articles are shown in
Fig. 14.8 . Figure 14.9 also shows the average and standard deviation of Q ( T ) and
QD( T ) for both featured and nonfeatured articles. Featured articles on average
contain high-quality content 86% of the time. Interestingly, this value increases
to 99% if we only consider the last 50 revisions of the articles. The same statistics
for nonfeatured articles show that they have high-quality content 74% of the time.
The difference between the averages of Q ( T ) and QD( T ) suggests that typically
low-quality content has short life span. This result is consistent with other studies
reporting the rapid elimination of vandalism in Wikipedia [ 23 , 45 , 46 ]. For exam-
ple, Kittur et al. [ 46 ] reported that about one third to one half of the systematically
inserted fictitious claims in Wikipedia are corrected within 48 h.
Search WWH ::




Custom Search