Online social decay

The information in this page is based on our work: A Theoretical Model for Understanding the Dynamics of Online Social Networks Decay” and “Stochastic Modeling of the Decay Dynamics of Online Social Networks

Online social decay is what happens when a social network suffers from the lack of activity causing a decay in its structure. Here we present visualization for some closed stack exchange websites that are close due to the inactivity of the their members. All data (including networks, gephi files, the GIFs and video files, and the code) are available upon request.

  • The Social network of Startup Business site over time. The whole data set was split into 47 snapshot with 15 days time interval.
    The decay of Startup Business website
    The decay of Startup Business website. Click to enlarge.

     

  • The Social network of Astronomy site over time. The whole data set was split into 10 snapshot with 30 days time window.
    The decay of the Astronomy website. Click to enlarge.
    The decay of the Astronomy website (1/2). Click to enlarge.

    The decay of the Astronomy website. Click to enlarge.
    The decay of the Astronomy website (2/2). Click to enlarge.

 

  • The Social network of Firearms site over time. The whole data set was split into 10 snapshot with  5 days time window.
The decay of the Firearms website (1/3). Click to enlarge.
The decay of the Firearms website (1/3). Click to enlarge.
 The decay of the Firearms website (2/3). Click to enlarge.
The decay of the Firearms website (2/3). Click to enlarge.
The decay of the Firearms website (3/3). Click to enlarge
The decay of the Firearms website (3/3). Click to enlarge

 

To better understand what happens during the decay process, we provided a model that represents the inactivity of the nodes. An illustration of the model using an example network. The color of the nodes represents how likely a node will leave in the future. White nodes are very unlikely to leave, Black nodes are the left nodes, and the nodes with colors in between white and black are more likely to leave depending on colors closeness to the black. Whenever a node leaves the network it is marked as black, all its edges were removed, and all of its neighbors gets affected by its leave by gaining more leave probability.

An example on how our model works.
An example on how our model works.

The mechanistic steps of the model are computed based on the following equation:

where the probability gain due to the leave of a node v is calculated by considering the leave probabilities of v‘s neighbors and the tie strength between v and its neighbors. More details can be found on the related paper.