Online conversation as networks
10 likes3,594 views
The document discusses the analysis of online conversations through network visualization, emphasizing the role of moderators in enhancing community engagement and communication. It presents metrics indicating that community-generated content is on the rise but highlights the importance of moderation to maintain interaction quality. Additionally, it examines the emergence of specialized subcommunities and the need for validation in discussions to ensure the richness of content.
Online conversation as networks
- 1. Online conversation as networks February 2014
- 2. Got data. Now what?
- 5. Overwhelmed?
- 6. Look closer!
- 7. What can you say about a conversation without reading it all?
- 8. What can you say about a conversation without reading it all? • look at the visualization
- 9. What can you say about a conversation without reading it all? • • look at the visualization guess what it means
- 10. What can you say about a conversation without reading it all? • • • look at the visualization guess what it means how would you, as community manager, react to this information?
- 11. What can you say about a conversation without reading it all? • • • • look at the visualization guess what it means how would you, as community manager, react to this information? “the vocalizer” of the network analysis software conveys interpretations for you
- 12. Very important!
- 13. Very important! • The network must be specified. It is important to define what nodes and edges represent.
- 14. Very important! • The network must be specified. It is important to define what nodes and edges represent. • Here (mostly): nodes => users, edges => comments
- 15. Is my team of moderators doing a good job?
- 16. With moderators Density 0.028 Average weighted degree 10.87 Average distance 2.3
- 17. No moderators Density 0.013 Average weighted degree 3.7 Average distance 3.3
- 18. With moderators Density 0.028 Average weighted degree 10.87 Average distance 2.3 No moderators Density 0.013 Average weighted degree 3.7 Average distance 3.3
- 19. Moderators rock!
- 20. Moderators rock! • “No user left behind”: they connect to the general conversation people that are not talking to each other.
- 21. Moderators rock! • “No user left behind”: they connect to the general conversation people that are not talking to each other. • They reduce the average distance between members by one hop
- 22. Moderators rock! • “No user left behind”: they connect to the general conversation people that are not talking to each other. • • They reduce the average distance between members by one hop They increase the average weighted degree (number of interactions on the network) from 4 to 11
- 23. How sustainable is my community?
- 24. moderators Community 400 2000 300 1500 200 1000 100 500 0 Nov Dec Jan Feb Mar Apr May Jun Jul Posts by contributor (cumulated) 0 Nov Dec Jan Feb Mar Apr May Jun Jul Comments by contributor (cumulated)
- 25. Share of Comments Share of Posts 1 0,75 0,5 0,25 0 Nov Dec Jan Feb Mar Apr May Jun User generated content as a share of the total content Jul
- 26. Share of participating users Share of relationships 1 0,75 0,5 0,25 0 Nov Dec Jan Feb Mar Apr May Jun Share of the conversation that does NOT involve moderators Jul
- 27. Not much, but getting better
- 28. Not much, but getting better • Almost all posts and about two thirds of the comments are produced by the community, not the moderators.
- 29. Not much, but getting better • Almost all posts and about two thirds of the comments are produced by the community, not the moderators. • BUT A lot of the interactions involve moderators. Drop them at your own risk!
- 30. Not much, but getting better • Almost all posts and about two thirds of the comments are produced by the community, not the moderators. • BUT A lot of the interactions involve moderators. Drop them at your own risk! • On the plus side, the share and number of interactions NOT involving moderators are both growing.
- 32. Moderators removed Modularity 0.5 (>0.4 = “high”) Subcommunities are clearly visible
- 33. Are subcommunities specializing? Bluer edge = more specialization on one or few topics, Redder edge = no specialization, all topics equally discussed
- 34. It seems so! (but check absolute numbers)
- 35. How “broad” and “deep” are my results?
- 36. Topic tree Blue research questions are addressed by green posts, in turn validated by red comments
- 38. Topic tree (detail) No validation
- 39. Topic tree (detail) Lots of validation
- 40. Topic tree (detail) Broad discussion, many points of view
- 41. Topic tree (detail) Narrow discussion
- 42. Some are, some are not...
- 43. Some are, some are not... • Check you are not relying on results from “narrow” and “shallow” discussion, however you might like the content!
- 44. Some are, some are not... • Check you are not relying on results from “narrow” and “shallow” discussion, however you might like the content! • Deploy moderators to ask the community to reinforce semmeingly important results that are not sufficiently validated
- 45. Now go do it!
- 46. A SNA module for Drupal
- 47. coming soon....