Discovery-Driven Graph Summarization英文学习资料 .pdfVIP

Discovery-Driven Graph Summarization Ning Zhang #1 , Yuanyuan Tian ∗2 , Jignesh M. Patel #3 # Computer Sciences Department, University of Wisconsin-Madison, USA 1nzhang@ 3 jignesh@ ∗IBM Almaden Research Center, USA 2ytian@ Abstract— Large graph datasets are ubiquitous in many do- relationships, and to control the resolutions of the resulting mains, including social networking and biology. Graph summa- summaries. By controlling resolutions, users can interactively rization techniques are crucial in such domains as they can assist drill-down or roll-up to navigate through summaries with in uncovering useful insights about the patterns hidden in the different resolutions. underlying data. One important type of graph summarization is to produce small and informative summaries based on user- Examples of the k-SNAP summaries is shown in Figure 1. selected node attributes and relationships, and allowing users to This summary is constructed over a coauthorship network interactively drill-down or roll-up to navigate through summaries (sketched in Figure 1b), where each node represents an author with different resolutions. However, two key components are and has a numerical attribute indicating the number of papers missing from the previous work in this area that limit the use of this method in practice. First, the previous work only published by the author. Figure 1a and Figure 1c show