Comments on: on visualizing online discussions

By: Tony Hirst

Tony Hirst — Wed, 21 Mar 2012 20:20:17 +0000

I’ve been pondering “accession” charts that I cribbed from @mediaczar – original examples at: http://blog.ouseful.info/2012/02/06/visualising-activity-round-a-twitter-hashtag-or-search-term-using-r/

x-axis is timestamp, y is unique user, though it could be thread? If y-axis is thread, colour or symbol (or colour/symbol pariings) could be used to denote different authors?

By: dnorman

dnorman — Tue, 20 Mar 2012 20:27:58 +0000

yeah. another layer of the visualization will represent the social/teaching/cognitive presence indicated in each post. not "quality" either, but gets to what happened, rather than straight wordcount.

the problems is I'll have so much data describing each post - the basic metadata, the coded presence data, conversation flow data, etc... that it'll be hard to visualize it meaningfully. non-trivial...

By: David Esrati

David Esrati — Tue, 20 Mar 2012 19:20:31 +0000

And, this is why we call it Piled high and deep.
Quality isn’t something that measures in word count or- numbers of comments- or anything else.
Did General McAuliffe need more words to respond to the Germans? “Nuts” did it.
Einstein said a lot with E=MCsquared.
I’d be more interested in how many people were engaged- and not how many commented.
Google has been using “reputation” for rankings- as well as numbers of visitors plus heuristic data for ranking for a long time.
Of course- all these charts and big words confuse me….
is the main reason the numbers of discussions on these platforms instead generated by a failure of the teacher to properly explain the subject- forcing the students to have to have a discussion?

By: dnorman

dnorman — Tue, 20 Mar 2012 05:23:29 +0000

I just added another mockup. This one includes the metadata from all of the Blackboard discussion boards in my data, as well as one of the 8 WordPress sites. This could get messy...

By: David Porter

David Porter — Mon, 19 Mar 2012 22:10:03 +0000

You might also want to look at Discovertext.com for its ability to thematically tag interactions based on transcation types that you define. Can ingest RSS, tweets, and even YouTube comments. Text is a breeze.

I see what you're trying to do by modeling the Rosling look. There's got to be a multi-dimensional modeling tool that could work with categories you define.

By: dnorman

dnorman — Mon, 19 Mar 2012 20:05:38 +0000

although, holy crap! that Network Analysis tool will come in handy for describing cohesiveness of a community on each platform. awesome!

By: dnorman

dnorman — Mon, 19 Mar 2012 20:04:23 +0000

The main reason was data compatibility - tools seem to work with one kind of data (LMS or blogs) but not both. I need to be able to compare activity in discussions taking place in both Bb8 and WordPress.

I'm downloading AGNA right now. Looks interesting. Thanks!

(downloaded...)

Looks like AGNA is mostly for visualizing connections within a social network - weighted links between individuals, rather than modelling the activity of a discussion. I'm looking for a way to represent the transactions of a discussion, to show patterns of activity.

By: David Porter

David Porter — Mon, 19 Mar 2012 19:57:31 +0000

Any reason you would not want to use some SNA tools to plot the conversation, theme or genre flows between the various participants, too?

Have you looked at some of the free SNA tools like Agna, D’Arcy? Not really sophisticated in presentation, but good at what it does.

http://mac.softpedia.com/get/Network-Admin/AGNA.shtml

By: dnorman

dnorman — Mon, 19 Mar 2012 18:18:31 +0000

yeah. of the 3 day-1 posts by the instructor, only 1 of the posts triggered responses. (and half of the responses to that post were by a student who didn't provide consent, so I had to use the null-post marker to at least make use of the metadata about their posts: date, position in thread)

By: Scott Leslie

Scott Leslie — Mon, 19 Mar 2012 16:47:00 +0000

I like the suggestion of using area rather than diameter to reflect post length as that will make it more manageable. Another idea - the width of threads doesn't need to be standard, does it, so they could adjust to the widest (longest) post.

My $0.02, your first effort is by far the best one. One of the main reasons I like it is that it reads downwards, like discussion forums do. Number 2 is less successful but still captures some of it, whereas for me, number 3 does not really work. Size and word count work well together visually, and this totally looses that.

Is diagram 3 correct - it seems to show one of the posts from day 1 leading to a discussion, whereas I read the other two visualizations as saying these did not spawn discussion.

I assume you are looking to do two visualizations, one for the CMS-based discussions and one for the blogs, or are both going to be done on the same graph?