Graphs of social networks have reached sizes of billions of edges. The scale of these graphs poses challenges in their analysis. Most graph algorithms are data driven and memory-based approaches usually do not scale because of the limit in capacity of single machines. In order to analyze the connectivity and the clustering coefficient of a huge graph, we need distributed and parallel approaches to handle the amount of data of these graphs. The goal of our research project was to find a framework that is able to handle the calculation of different graph properties.
Carina Saal, Pavithran Sakamuri, Tung-Yin Kuo, Roman Sokolov
15.10.2015 - 10:15