The SSNA of the MENA Open Village conversation!


The Social Semantic Network Analysis of the Open Village conversation is finally here!

  • 147 participants in the conversation
  • 1,568 posts in 243 topics
  • coded with 914 ethnographic codes in 2,526 annotations
  • the dashboard is here

The code co-occurrence graph is, as always, very large, and not readable by humans simply via visual inspection: it has 914 nodes and over 10,000 edges. I downloaded the graph and tweaked it a bit in search of a high-level view on the conversation. I ended up producing this image:

  • Only edges that represents at least 4 co-occurrences are shown. This means that the two ethnographic codes connected by each of these edges appear together on at least 4 posts. The maximum number of co-occurrences is 10.
  • The graph has 35 nodes and 79 edges
  • Darker edges represent more co-occurrences.
  • Node colors represent “clusters”; or to be more precise, the classes of the modularity-maximizing partition. This network is not highly modular (about 0.23), but you can see that blue nodes seem to be more or less “techniques”, “way of working”, or “moves” that young people in MENA can apply to their predicament, whereas the green ones represent, in many cases, “solutions” like co-working, co-living and entrepreneurship. There is some intuition, I think, to be gained by looking at the yellow nodes (around community health) and the orange ones (around improve community).

Any thoughts, anybody?

MENA Open Village Coding: Lessons Learned

While working on it, I noticed that “obstacles” might be a pretty interesting node to explore. Centering in on that on both what it co-occurs with and what its second degree connections are might at least be an interesting topic for a blog post. It’s a clear subject line too - experienced obstacles to social entrepreneurship among the Open Village participants.


Good idea, @hugi. Unfortunately, when I tried to dig deep into the data (reading the actual posts) around obstacles, they come out empty:

This was obtained as follows: from “code view full”, click on the eye interactor; next, click on an edge. Expected behavior is to show you a list of posts that were coded with both the codes connected by that edge.

This is Chrome, not Firefox. Obviously this bug is not related to the network’s layout. :frowning:

Setting up Graphryder API with new dataset

Oh dear. Not good. I’ll try to have a quick look today, but I’m really swamped. Admittedly, I didn’t test the dashboard a lot since I was caught up with the bugs in the API.


@khaoula Voici une démonstration concrète des travaux de Alberto menés sur la platforme dont je t’ai parlé quand on a échangé durant notre rencontre à El space


Thanks @zmorda . What about deep understanding of these semantic representations? @alberto we could understand further what is mean this word?