%0 Journal Article %1 Hill:September_2006:1061-8600:584 %A Hill, Shawndra %A Agarwal, Deepak K. %A Bell, Robert %A Volinsky, Chris %D 2006 %J Journal of Computational & Graphical Statistics %K data network representation toread %P 584-608(25) %R doi:10.1198/106186006X139162 %T Building an Effective Representation for Dynamic Networks %U http://www.ingentaconnect.com/content/asa/jcgs/2006/00000015/00000003/art00006 %V 15 %X A dynamic network is a special type of network composed of connected transactors which have repeated evolving interaction. Data on large dynamic networks such as telecommunications networks and the Internet are pervasive. However, representing dynamic networks in a manner that is conducive to efficient large-scale analysis is a challenge. In this article, we represent dynamic graphs using a data structure introduced in an earlier article. We advocate their representation because it accounts for the evolution of relationships between transactors through time, mitigates noise at the local transactor level, and allows for the removal of stale relationships. Our work improves on their heuristic arguments by formalizing the representation with three tunable parameters. In doing this, we develop a generic framework for evaluating and tuning any dynamic graph. We show that the storage saving approximations involved in the representation do not affect predictive performance, and typically improve it. We motivate our approach using a fraud detection example from the telecommunications industry, and demonstrate that we can outperform published results on the fraud detection task. In addition, we present a preliminary analysis on Web logs and e-mail networks.

@article{Hill:September_2006:1061-8600:584, abstract = {A dynamic network is a special type of network composed of connected transactors which have repeated evolving interaction. Data on large dynamic networks such as telecommunications networks and the Internet are pervasive. However, representing dynamic networks in a manner that is conducive to efficient large-scale analysis is a challenge. In this article, we represent dynamic graphs using a data structure introduced in an earlier article. We advocate their representation because it accounts for the evolution of relationships between transactors through time, mitigates noise at the local transactor level, and allows for the removal of stale relationships. Our work improves on their heuristic arguments by formalizing the representation with three tunable parameters. In doing this, we develop a generic framework for evaluating and tuning any dynamic graph. We show that the storage saving approximations involved in the representation do not affect predictive performance, and typically improve it. We motivate our approach using a fraud detection example from the telecommunications industry, and demonstrate that we can outperform published results on the fraud detection task. In addition, we present a preliminary analysis on Web logs and e-mail networks. }, added-at = {2011-02-09T15:33:32.000+0100}, author = {Hill, Shawndra and Agarwal, Deepak K. and Bell, Robert and Volinsky, Chris}, biburl = {https://www.bibsonomy.org/bibtex/2c4c90214919c4edb8da5d69b78e5180b/hotho}, doi = {doi:10.1198/106186006X139162}, interhash = {0bef8e24366140d674636ff4f032a8de}, intrahash = {c4c90214919c4edb8da5d69b78e5180b}, journal = {Journal of Computational & Graphical Statistics}, keywords = {data network representation toread}, month = {September }, pages = {584-608(25)}, timestamp = {2011-02-09T15:33:33.000+0100}, title = {Building an Effective Representation for Dynamic Networks}, url = {http://www.ingentaconnect.com/content/asa/jcgs/2006/00000015/00000003/art00006}, volume = 15, year = 2006 }