Professor and Ph.D. student's paper published in PLOS ONE


April 9, 2015

ECE Professor Radu Marculescu and Ph.D. student Huan-Kai Peng's paper, Multi-Scale Compositionality: Identifying the Compositional Structures of Social Dynamics Using Deep Learning, has been published in PLOS ONE. PLOS ONE features reports of original research from all disciplines within science and medicine. By not excluding papers on the basis of subject area, PLOS ONE facilitates the discovery of the connections between papers whether within or between disciplines.

Paper summary
Everything in social media evolves with time; understanding the patterns of this change is essential for any social-media application. The SLD research group at CMU has recently proposed a deep-learning method that can efficiently mine social media and identify such patterns of social dynamics, like in Twitter or Yelp datasets. This way, it becomes possible to identify compositional structures at multiple time scales, some lasting only minutes, while others being present for days. This approach is a key enabler not only for understanding and forecasting the social-media dynamics, but also for engineering social media dynamics, with direct applications to opinion spreading and online contents promotion.

View the paper here.