Automatic Text Analysis of Values in the Enron Email Dataset: Clustering a Social Network Using the Value Patterns of Actors

This paper describes an automatic text analysis of values contained in the Enron email dataset that seeks to explore the potential to apply value patterns to cluster a social network. Two hypotheses are posed: individuals communicate more frequently with other individuals who share similar value patterns than with individuals with different value patterns; and people who communicate more frequently with each other share similar value patterns. The first hypothesis is supported: indeed, individuals were found to communicate more frequently with individuals who share similar value patterns, and further, the extent to which this is true appears to depend at least in part on the value patterns themselves. However, the second hypothesis is not supported - people who communicate more frequently with each other do not necessarily all fit into a particular value type. Thus, values have utility as a novel tool for social network analysis.