Summarization and Information Loss in Network Analysis∗

The increasing availability of large-scale network data makes the problem of network summarization especially relevant. In any data summarization, however, it is important to remain aware of the information not being presented. As such, we present a mathematical framework within which to consider the problem of network summarization. Using that framework, the concept of information entropy is applied in the context of network summarization. Finally, an O(—E—) algorithm is presented for computing information entropy in the case of simple deterministic network summarizations.