This document proposes and evaluates distance measures for dynamic citation networks. It discusses how standard community detection methods do not capture the expected stability of relationships in citation networks. The authors introduce a distance measure based on the idea that two documents are more similar if they share more common "sinks", or ideas introduced into the network. They test this sink-based distance measure through simulations and on a United States Supreme Court citation network, finding it performs better than standard approaches at revealing stable communities over time.
Related topics: