The Network Analysis graph reveals conversation and mention patterns between the most central Twitter profiles of a community or healthcare topic.

The two main elements of the graph are nodes and edges.

Nodes (the circles in the graph) represent Twitter profiles in a community and are initially color-coded by its assigned stakeholder category color if any.  The size of the node is also significant as it represents the size of the profile's influence within this community.

Edges (the lines that connect the nodes) represent the communication between the two connected nodes with the arrow indicating the direction of the communication. The edge type could either be a reply, a mention, a retweet, or a quote tweet.  The edges are also distinguished by its thickness; thicker edges indicate more communication between the two profiles (or nodes).

 

Engagements

  • Filter by stakeholder category.  Network graph can be filtered to display nodes and edges of the stakeholder categories selected from the stakeholder menu.
  • Graph view control.  Zoom in and out of the graph using the buttons in the toolbar.  Click and drag the canvas to move the entire network graph.  Click and drag a node to new location on the canvas, or to create more separation.
  • Profile details and profile neighbors. Click on any node to open popover with additional details of the Twitter profile with options to view its Healthcare Social Graph page, Tweet Transcript, or Interactions.  Hover over any node to view its direct neighbors (nodes and edges) in the network graph.
  • Graph settings controls . Click on the settings button in the tool bar to access graph settings.

Profiles tab

  • Max number of profiles to display.  The slider changes the limit amount of the query to the database for the maximum number of profiles to retrieve.  Default is 100, and the range is from 20 to 500.
  • Show profiles.  The slider that filters (or hides) nodes on the graph by size. 
  • Profile labels.  Slider that controls the amount of profile labels to display on the graph.
  • Show links.  Slider that filters (or hides) the links between the nodes on the graph by thickness
  • Graph layout options.   Click on one of the buttons to trigger the animation to change the network graph to a new layout.  Five layout options are currently provided at this time: circular, force 1, force 2, force 3, and random.

Communities tab

The Louvain algorithm is used to detect communities in the network graph.  Turning on the community detection, each community will be assigned a color.   Switch to different color set by clicking on "Randomize color" button until you find a color that is preferable.

A table lists all the communities from the largest to the smallest with its assigned color, the name of the most influential profile of that community, and the size of the community.  Mouse over any row in the table to highlight the selected community on the network graph.  

Data tab

This tab contains settings to change the data source used for the network graph.

Tweet type

  • All.  The connection is either a mention, a reply, or a quote tweet.
  • Mentions.  Only mentions between accounts are drawn.
  • Replies.  Only replies between accounts are drawn.
  • Quote Tweets.  Only quote tweets between accounts are drawn.

Tweet direction

  • All.  Connections are all unidirectional with the arrowhead showing the direction of the connection.
  • Bidirectional.  Displays connections where both users are mutually mentioned, replied, and/or quoted each other.

Rank by

  • SymplurRank.  The nodes use our SymplurRank algorithm to return most influential profiles for the given dataset, we have found this to be more effective at filtering out retweet bots, spam accounts, and interactions that may not be relevant.
  • Count.  Does not use our SymplurRank algorithm but simply the tweet count as a criteria for returning influential profiles.

Further Notes

  • Node Magnitudes: If Rank By is SymplurRank then the Node Magnitude numbers equals the SymplurRank algorithm's outputs. If Rank By is Count, then Node Magnitude is the number of edges that are connected to the node.
  • Edge Magnitudes: This is based on the magnitudes of the nodes that the edge is connecting: edge mag = user_a_node_mag * user_b_node_mag * repetitions
Did this answer your question?