I’m thinking about using Discourse in an educational setting and wonder what data is available about posts. The data will help me design analytics to understand student engagement and understanding.
Hi John, do you mean what stats are available? If you install data explorer you can access the entire database.
Is there a list of all of the data items?
The data explorer plugin includes all of the tables. Also, discourse is open source, so it’s all at github. You can look in
/app/models and each file includes the associated table at the bottom.
Thanks. Those models spelled it out nicely. I see that there are user-defined fields, but they look like they extend the user profile. If I wanted to create custom post-level data (for analytics), how difficult would it be to do the calculations and them to another table in Discourse or another database entirely?
You can customise Discourse to your hearts content with plugins, the main constraints being time and money (though some things are harder than others of course)
However, have you thought about coming at this from the other direction?
What do you actually want?
There are quite a lot of analytics at your disposal out of the box. Something might already exist that in some way meets your requirement …
Also make sure you make the explicit distinction between Topics and Posts in whatever you specify.
It sounds like you are making trouble. Plugins can add post and topic custom fields.
Indeed. I use a lot of EdTech tools and it would be nice if they were written on top of existing, mainstream products. In this educational use case, schools need SSO and a way to pull class rosters to set up discussion groups. Discussions are underutilized in education because they lack analytics teachers need to easily monitor and advance discussions. They are also underutilized because most teachers haven’t used it enough to become proficient. In all of my grad classes, the online discussions were poorly moderated and viewed as a to-do item instead of intellectual discourse. The growth of virtual classrooms and speect-to-text tools make it an even more attractive platform for schools.
Yup. Back when I was an open source software zealot and an idealistic freshly minted PhD from Top Tier university working at Large Public University in the Southeast, I got great support from the tech folks when I wanted to to hook into their SSO when I used Moodle (rather than the university-supplied Blackboard). At the small 3rd-tier university where I started using Discourse, I went to some length to keep the tech folks from even knowing about it.
My colleagues (and now my wife, who’s now tenure track at a Flagship in the west) who studied conversation analysis wouldn’t even consider trying to do anything other than use the prescribed tools. Even with SSO in place, it’s a tough transition.
If you have a budget, or maybe a Discourse-vs-whatever platform study that my wife could co-author, I might be able to help beyond what you can get here. The SSO stuff is pretty well developed now, so if you get even minimal help from the folks who hold the keys to the kingdom, it shouldn’t be that hard, and if your instition uses Google or Office365, you can follow existing howtos and not even tell them.
I have budgetary support. If the addon could be a viable EdTech tool, I’m happy to fund it myself.
I like Jay’s approach. Do it and ask for forgiveness later. Online community is hard to do well. Many thankless hours of tending, begging, cultivating and growing one person at a time.
People don’t like to write well. Social media has taught us to write poorly and sometimes even lie. It’s a bad state of affairs that I hope educational institutions can address through positive online experiences.
Completely agree to this statement.
I am also trying to integrate Discourse for an Online program and we use GSuite. The SSO integration was pretty straight forward (after I saw the post).
I found the Data Explorer plugin to be a very useful tool for initial reports. With a good SQL programmer, I am able to get pretty much all the main reports that I want (especially when I want to filter for specific categories and do multiple join statements). We have just started our exploration, but already Discourse is offering so much that it will take us some time to explore them to the fullest. The network graph is one addition that I would like to suggest as a possible add-on feature/plugin for educational institutions.
You mean like Community Network Visualisation?
I have a 6600+ user community, and as I understand from the post, this might cause performance issues while running the query.
I can’t use trust level filter as of now as all my learners will be in tl0 now (and possibly few in tl1 by end of this academic term). Any suggestions on optimizing for my use case?
It might, it might not, you’d have to test it out and see the job run at least twice and see how long it runs in Sidekiq, check the CPU and disk IO etc and usability whilst it’s running.
My only comment would be that 6600 users on a graph at once is going to be very noisy, so yeah, maybe consider using with the TL filter once you’ve got a more mature community that’s climbed the Trust Levels …?
Btw, questions like these are best asked on the plugin topic.
Yes. We are just starting on Discourse and we expect a steep learning curve on community building in the first few months.
Thank you for the pointer. Will keep this in mind. Just starting off in the community and slowly getting used to the structure. I have bookmarked the post in #plugin. I’ll update my results there after the first run.
No issues, just that other users of the plugin might benefit from your questions and use case.
Unless you have dramatically charged the defaults that does not seem right. It doesn’t take much to get to trust level 1, and I would expect most to be at tl2 in a week or two.
Yes, I’ve made some significant changes to the Trust Level settings. This ensures that serious participants from tl0 moves to tl1 early while non-active will take more time to get promoted.
@pfaffman : Just wanted to check something related to post limits to tl0 level.
I had setup higher thresholds for tl1 promotion (topics entered = 50, posts read = 300, time spent in mins = 500) in order to be very sure of the learners involvement in the forum before allowing more permission. But from some of the responses that I have been receiving from learners in the forum, I believe the first_day limit is still getting applied each day. Is there an implicit assumption that learners will progress to tl1 fast and hence such a usecase may not occur?