Research on Online Digital Cultures - Community Extraction and Analysis by Markov and k-Means Clustering

We investigate approaches to personal data analytics that involves the participation of all actors in our shared digital culture. We analyse their communities by identifying and clustering social relations using mobile and social media data. The work is part of our effort to develop tools to create a “social data commons”, an open research environment that will share innovative tools and data sets to researchers interested in accessing the data that surrounds the production and circulation of digital culture and their actors. This experiment focuses on the groups of clustered relations that are formed within a user’s social data traces. Community extraction is a popular part of the analysis of social data. We have applied the technique of Markov Clustering to the Twitter networks of social actors. Qualitatively, we demonstrate that it is more effective than the Louvain method for finding social groups known to the subjects, while still being very simple to implement. We also demonstrate that traces of cell towers captured using our “MobileMiner” mobile application are sufficient to capture significant details about their social relations by the simple application of k-means.