Using the EdgeSense demo

Navigate to http://wikitalia.github.io/edgesense/demo/. The page takes about 10 seconds to load.

Interpreting the network

• The network represent the online conversation in an online community called Edgeryders in late 2011 and 2012. Nodes represent users; edges represent comments (learn more about network representation in EdgeSense).
• If two nodes have the same color, they are part of the same subcommunity. By this term we mean a set of nodes that are more tightly connected to each other than they are to the rest of the network. Edges take the colour of their source node. Subcommunities are computed by applying the Louvain algorithm to the network. Nodes representing community administrators are coloured in grey, but the edges departing from them maintain the color of the subcommunities that administrators are assigned to.

Interpreting the metrics

At the top right of the page EdgeSense displays network statistics.

• Number of nodes. Equivalent to the number of users that have written or received at least one comment. It does not coincide with the number of "active users" as defined by Drupal; in most communities the number of nodes is significantly lower than the number of active users. Edgeryders had 1200 active users (in the Drupal sense) at the end of 2012, but only 260 nodes.
• Number of edges. Equivalent to the number of relationships established across users in the community. It does not coincide with the number of comments: if user i comments user j's content 4 times, the database records 4 comments, while EdgeSense aggregates them into one relationship of weight 4. In the demo data we have 4,100 comments aggregated into 1,600 relationships.
• Average degree. The average over the sum total of all relationships (generated both from incoming and outgoing comments) across all nodes. You can interpret it as the number of interactions that the average user has had in the community in the period considered. The higher this number, the more lively the community feels to the average user.
• Modularity. It measures the distance between the network and a random network with the same degree distribution, on a 0-1 scale. A value close to 0 means that there are no clearly defined subcommunities (or conversation groups) in the network. High values modularity (say over 0.3) might indicate the emergence of specialization, or "ganging up". Moderators often have the effect of lowering modularity, as they talk to many people and therefore pull the conversation together.

###Interpreting the timelines Below the network representation, EdgeSense visualizes three timelines.

• The first one shows posts per unit of time, divided by authorship (moderators vs. non-moderators: EdgeSense counts as moderators users that have roles other than "active user" in the database).
• The second one shows the same thing for comments.
• The third one shows the share of posts and comments written by non-moderators – that's your share of user generated content. The closer it is to 1, the less the conversation is dependent on moderators.

###Interacting with EdgeSense #

• Use the slider on top to "travel through time". Metrics are recomputed and referred to the time period you have selected. Nodes and edges appear and disappear according to their creation date. The network layout and the color-coding, however, are not recomputed and refer to the network at the end of the period. This allows users to preserve the visual identity of nodes and edges and follow the growth of the network over time.
• Tick the "moderators" box to remove community moderators from the network. Use this command to visualize the contribution to the conversation of the team of moderators; all network statistics update according to the selection.
• Hover over nodes to visualize user information. Our demo is anonymized, so you only see user ID.
• Click on the Lock icon to enable zoom-and-drag on the network. This allows you to explore in more detail different areas of the network.
• With the Lock icon on open, click on a node to display information about it. Information includes the user's name (clickable to access that user's profile page), in- and out-degree and betweenness centrality.
• With the Lock icon on open, double-click on a node to display her ego network, i.e. her neighbours and the edges connecting them to each other.
• Click on the Filter icon to access a pop-up selector where you can filter out subcommunities. This is useful to understand how different subcommunities interact with each other, or to explore the growth over time of subcommunities.