Implementation of the GraphScope algorithm for clustering evolving graphs. Detailed description of the algorithm is available on: http://dl.acm.org/citation.cfm?id=1281266. The graphs are stored in matrices so the implementation uses the EJML library (https://code.google.com/p/efficient-java-matrix-library/) to handle the required calculations.
Data folder contains the synthetic and real datasets used to test the algorithm. The datasets are described briefly in the following lines.
- data-synth folder contains four synthetic datasets. Each dataset contains 10 time step matrices, containing 1,000 objects and 1,500 features, assigned to 8 embedded clusters of objects and features. The dataset are available online (http://mlg.ucd.ie/dscc.html). More details can be found in the readme file inside data-synth folder and in the paper that describes the datasets (http://www.csi.ucd.ie/files/ucd-csi-2011-08.pdf). It's worth to mention that for the algorithm evaluation has been used static graphs also. The BenchmarkGenerator class contains three implementations of static binary graph benchmarks.
- delicious.2006.summer and delicious.2007.autum contain the real data used for the algorithm evaluation. The data correspond to the bookmark-tag correlation that has been recorded during the summer of 2006 and autum of 2007. Each snapshot contains 1,500 tags and 3,000 bookmarks and coves a 7-day period. These data come from the delicious dataset descibed here :http://www.dai-labor.de/en/publication/359 . Note that we keep the 1,500 most frequent tags and 3,000 most frequent bookmarks.
For more information or support, contact: sot.beis@gmail.com