Trust in Online Collaborative IS - Community-Built Databases: Research and Development - page 351

Fig. 14.3 Distribution of reputation for all users in English Wikipedia based on the three Models.
The X -axis shows the reputation bins and the Y -axis shows the percentage of users in the bins
to January 1, 2007, for reputation estimation and then examine users' behavioral
indicators on January 1, 2007, and September 30, 2009.
For each model, we classify all the users into ten different bins (ignoring bots)
according to their reputation values. For each bin associated with model, we cal-
culate the mean of four individual, time-dependent, behavioral indicators, namely
RDR, DSR, SDR, and CDR, defined as follows:
l Reverted Data Ratio (RDR) is the ratio of the number of submitted revisions by a
user that are reverted by other users to the total number of revisions submitted by
the same user. This metric can be interpreted as the tendency of a user toward
contributing vandalistic/problematic content.
l Data Stability Ratio (DSR) is the percentage of contributed data by a user who
remains live in the wiki pages. It shows the percentage of content contributed by
a user that has not yet been deleted by other users yet.
l Submission Data Ratio (SDR) is the number of revisions submitted by a user to
the total number of submitted revisions. This metric shows how actively each
user contributes to the wiki pages by submitting new revisions.
l Correction Data Ratio (CDR) is the ratio of the number of reverts done by a user
to the total number of reverts. This metric can be interpreted as the tendency of a
user to make corrections in the wiki pages.
Figure 14.4 shows the mean of CDR, SDR, DSR, and RDR, respectively, in each
bin associated with each reputation model, when the behavioral indicators and
reputation values are calculated using data up to January 1, 2007. As the diagrams
show, in general, there is a positive correlation between user reputation and CDR -
signifying that users with an estimated high reputation tend to make more correc-
tions than users with estimated low reputation. The positive correlation between
user reputation and SDR also shows that higher reputation users submit more
revisions compared to lower reputation users. The correlation between user reputa-
tion and RDR is negative, indicating that lower reputation users tend to contribute
vandalistic or low-quality content more frequently. These positive and negative