Analysing the discourse in discussion forums of CS projects can help us to understand underlying patterns of collaborative knowledge creation and to identify highly engaged users. We ask the question: What can the quantitative analysis of forum data tell us about these patterns?
A minority of power users is responsible for most of the discourse.
Volunteers shape the discourse to a larger extent than anticipated.
We analysed the discussion data of the “Talk” pages of seven CS projects with similar forums on the Zooniverse platform. One point was to distinguish the share of different user groups in the discourse and in collaborative knowledge creation. In the discussion forums, users are assigned to different user roles, specifically volunteers, scientists, and moderators. They can also be promoted to higher roles, i.e., volunteers can become volunteer-moderators. In addition to examining which users make most of the comments, we also considered these specific user roles.
The user roles are publicly indicated on the platform, and in total, volunteers are responsible for 75% of the comments, moderators for 23% and scientists for 2%. The moderator user group can be further broken down into moderators who were recruited from the volunteer group (49% of total cases), those who are also professional scientists (6% of total cases), and those without any further specification (i.e., other moderators, 45% of total cases). In Fig. 1 we can see that the top 5% of users (arranged by individual comments made) across our seven analysed projects are responsible for 75% of the total comments. In other words, the majority of users only marginally influence discourse in the forums. On the right we can see the distribution of user roles among these top users. About 70% are volunteers (i.e., regular citizen scientists), whereas 30% are moderators. These top moderators are also roughly composed of equal parts other moderators and promoted moderators, with another 5% originally being scientists.
The above figure shows a combination of multiple bar charts. The bar on the left shows the distribution of total comments across our seven analysed projects. In the middle, we see the top users, to the right the distribution of user roles among them and on the extreme right their composition.