Wikipedia graph mining: dynamic structure of collective memory

ABSTRACT Wikipedia is the biggest ever created encyclopedia and the fifth most visited website in the world. Tens of millions of people surf it every day, seeking answers to various questions. Collective user activity on the pages leaves publicly available footprints of human behavior, making Wikipedia a great source of the data for largescale analysis of collective dynamical patterns. The dynamic nature of the Wikipedia graph is the main challenge for the analysis. In this work, we propose a graph-based dynamical pattern extraction model, inspired by the Hebbian learning theory. We focus on data-streams with underlying graph structure and perform several large-scale experiments on the Wikipedia visitor activity data. We extract dynamical patterns of collective activity and show that they correspond to meaningful clusters of associated events, reflected in the Wikipedia articles. We demonstrate evolutionary dynamics of the graphs over time to highlight changing nature of visitors’ interests. Apart from that, we discuss clusters of events that model collective recall process and represent collective memories – common memories shared by a group of people. In the experiments, we show that the presented model is scalable in terms of time-series length and graph density, providing a distributed implementation of the proposed algorithm.