Department or Program



In exploring the evolution of networks, many researchers have focused on analyzing dynamics of unipartite networks. However, bipartite network structures are also essential in understanding complex co-evolutionary processes. A significant proportion of foundational work in bipartite networks has centered on ecological and biological networks as a way of understanding mutualistic interactions within communities. In this study, we use a in silico evolutionary model, inspired by a plant-pollinator communities to generate bipartite network data for a co-evolving resource-user system. We then analyze this data to understand network change over time. This thesis primarily focus on the visualization and mining techniques of key bipartite network structure indicators over time. Our goal is to understand how the structure of the bipartite network, as well as its projected unipartite graphs, changes over time. To have a better sense of the structural properties and key information we want to highlight, we employ several visualization techniques in representing our evolving objects. Given the fact that some statistics may have similar but different timings in which shifts appear, we then apply the model-free anomaly detection methods by Benkő et al (2022) based on calculations of the Temporal Outlier Factor (TOF) to these time-series indicators. This method automates the process of identifying the unique points/periods where qualitative changes may happen, so that we can compare statistics in the same network and also across samples. Although detection results signal the period of transition on the evolutionary process, model-free algorithms can be less effective when parameters like expected event length is undetermined.

Level of Access

Open Access

First Advisor

Diaz Eaton, Carrie

Date of Graduation


Degree Name

Bachelor of Arts

Number of Pages


Components of Thesis

1 pdf file

Open Access

Available to all.