Analyzing Stanford’s Academic Network

1. INTRODUCTION The university is a fundamental social institution. It is a haven for research and is where the next generation gets educated. Gaining knowledge of its properties and understanding how it functions are thus of inherent value to society. In our project, we analyzed Stanford's academic network , where faculty members are vertices and edges between faculty members represent academic relationships like publication co-authorship, being on the same dissertation committee , and being on the same grant. Our motivation for studying this particular network, aside from its inherent social value, is that it has several interesting features from a network analysis standpoint. Firstly, it is the prototypical example of a hierarchical network. Secondly , there are many different relationships between people in the network; essentially there are many different networks that co-exist on the same underlying vertex set. Finally, our data has a strong temporal element since we have yearly snapshots of each network over the period from 1994-2007. Our project is divided into three main goals. The first goal is to discover the community structure in the network. Its hierarchical structure is well-known, but how well does the activity of the network conform to the university's official divisions? Are there more coherent groups than the fac-ulties? We applied graph clustering techniques to separate the network into natural clusters. With this partitioning, we quantify how isolated or connected certain fields of study are from the others, and how weak or strong interdisciplinary collaboration is between specific departments. The temporal nature of our data made this even more interesting; we quantify how collaboration varies over time. Our second goal is to verify if Burt's theory of structural holes apply to the university setting. Are faculty who fill structural holes more successful researchers? Finally, the third goal is to briefly explore if we can leverage the many different edge types we have to predict grant proposal success. The theory of structural holes was pioneered by Ron Burt [1, 4, 3]. He found that people who fill structural holes in networks derive social capital from their advantageous position. In his work, Burt proposes " network constraint measures " to detect hole signatures and thus identify structural holes. His studies were done in commercial settings (in [3], he analyzed a large electronics company), with information gathered from questionnaires. In this paper, he gave evidence that people who fill structural holes accrue various benefits …