NodeXL Provides Complex Visualizations of Interconnected Facebook Likes and Comments

Graph taken from NodeXL gallery

NodeXL is a data visualization tool that is capable of producing graphics that represent networks of social media interactions. This technology relies on the graph data structure used in computer science to represent data points (called "nodes" or "vertices") and the pathways that connect them (called "edges"). By creating graphs, we can determine how different data points are connected to each other and notice patterns between different areas of the graph.

For instance, the graph shown here is a depiction of comments and likes of posts on the Facebook page for "MarketingProfs." The edges, marked in green, show relationships between individuals who have interacted with the page. Edges are drawn between two users who comment consecutively, two users who like the same post, or a user who likes or comments on a post and the post's author.

Unfortunately, the graph format is not easily decipherable to an audience who is not previously familiar with the data gathered. The graph is divided into 6 clusters labelled G1 through G6, and the reason for these divisions is the "Clauset-Newman-Moore cluster algorithm," which is not a term understood widely by a general population. Furthermore, the graph relies on "centrality values" and "edge weight" values to determine different attributes such as edge thickness and vertex size, but does not indicate how these values are obtained. To make this graph more accessible to the general public, I think that the graph's creator should either elaborate on how these values are related to the comments and likes in the network, or perhaps color-code the edges based on the relation between vertices (e.g. commenter to author, two commenters, etc.).

I think that NodeXL provides a great way to discover connections in social network data, especially in our society where a high amount of pressure is placed on companies to analyze their clients' patterns. Unfortunately, the usefulness of the data is occluded by a confusing model of presentation. With a clear legend and more obvious description of how data points are clustered, I feel that this visualization could better portray the patterns found in these social media interactions.