Looking at a Network: From the Inside Out
Examining networks is obviously the whole point of this class but a side effect of this is we have to have the entire graph to draw conclusions and do the types of analysis the homework and tests require. However, when the entire network is unavailable to be viewed the only option left is to look to study the system from the point of view of a node instead of an omniscient network god. This article I found looks to do just that by visualizing the LinkedIn network structure but from the point of view of the individual user.
Using python in Jupyter Notebooks, Tavish Gobindram looked to apply data visualization techniques to create an interactive and colorful representation of his personal set of LinkedIn connections. His motivation was just to have a better grasp of his connections, which companies represent which portions of his network but from this effort, we are able to look at the unique perspective a node has in a graph.
LinkedIn being an obviously very large network, it would be impossible to analyze by hand even a small subset of the graph however using Tavish’s network as an example we can analyze what a node “sees” and verify principles we have learned in class. One good example of this is the principle that nodes in a graph that have a common first-degree neighbor are likely to become friends. We can see this in the visualization with a large number of J.P. Morgan connections which will only grow and friends of friends become connections over time. Another topic we can see the potential of is the high probability of a local weak bridge forming. Although connections are not ranked by strong vs weak on LinkedIn we can still assume that singular connections to a company would imply that connection to be a weak connection. We can also infer that that weak connection leads to a user who has a large component of strong connections just like Tavish does with J.P. Morgan.
Although this analysis falls apart in corner cases such as with my LinkedIn being connected to a majority of Cornell students and not many company employees, the analysis of a network from the perspective of a single node offers an interesting perspective on the topics and principles we discuss in class.
Sources:
https://towardsdatascience.com/visualizing-my-linkedin-network-c4b232ab2ad0