Data Inovasi Perguruan Tinggi
Universitas Telkom - Artikel Hasil Penelitian
Graph based modelling is common in many implementation areas involving combinatorics relationship such as in social network. The data explosion produced from user generated content in online social network services trigger the emergence of large-scale social network. Having large graph at our disposal gives us many opportunity but at the same time increase the complexity problem, especially in several graph metric computations and also at graph visualization. A fast summarization methods is needed to reduce the graph size into the only most important pattern. This summarize sub-graph should represent the property or at least converge to the value of the original graph property. Social Network is characterized by scale free degree distributions, which have fat-head less important nodes that can be removed. Graph Pruning method is introduced to remove less important nodes in certain graph context, thus reduce the complexity of large-scale social network while still retain the original graph properties. The method is based on k-core graph properties.??The paper shows how is the effect of graph pruning to the several most used social network properties.