Digital News Graph Clustering using Chinese Whispers Algorithm

As the exponential growth of news creation on the internet, the amount of digital news has reached out billion numbers. Digital news is naturally linked each other but it needs to be grouped so that user can easily classify the news that they read. Graph is the most suitable data model to represent digital news since its can describing relation in easy and flexible manner. Thus, to overcome grouping problems, in this paper we using Chinese Whispers Algorithm as the graph clustering approach. We choose Chinese Whisper Algorithm based on consideration that the algorithm is able to make clusters from a big graph data with a relatively fast process [8], that appropriate with the characteristics of digital news. In this research, we examine the graph quality by comparing intra and inter-cluster weights of every node. This scenario gives us a quite high result that 95% of nodes have intra-cluster weight higher than its inter-cluster weight. We also investigate the graph accuracy by comparing the cluster results with expert judgement. As the result, the average accuracy of digital news graph clustering using Chinese Whisper algorithm is 80%.