Storyline detection and tracking using Dynamic Latent Dirichlet Allocation
Daniel Brüggermann, Yannik Hermey, Carsten Orth, Darius Schneider, Stefan Selzer, Gerasimos Spanakis · 2016
In this paper we consider the problem of detecting and tracking storylines over time using news text corpora.World wide web creates vast amounts of information and handling, managing and utilizing this information is difficult without having systems that are able to identify trends, arcs and stories and how they evolve through time.The proposed approach utilizes a dynamic version of Latent Dirichlet Allocation (DLDA) over discrete time steps and makes it possible to identify topics within storylines as they appear and track them through time.Moreover, a graphical tool for visualizing topics and changes is implemented and allows for easy navigation through the topics and their corresponding documents.Experimental analysis on Reuters RCV1 corpus reveals that the proposed approach can be effectively used as a tool for identifying turning points in storylines and their evolutions while at the same time allowing for an efficient visualization.