Representing and modeling information has become a pivotal part of modern science. Most of the interesting problems today originates from complex systems that are not isolable neither dependent solely on a single discipline. In fact, for this kind of problems, it has become hard to even guess where a discipline ends and another starts. A contributing factor to this trend is the need of methods to analyze, characterize, organize and visualize information. Such tasks now span along several disciplines, including physics, statistics and computer science. In such a complex environment, network science emerged as a suitable and general representation for a myriad of real-world systems. This includes representing information itself in terms of its intricate relationships. However, since there is still a lack of such techniques, probing their structure and dynamics is still a challenging problem. In this proposal, we intent to continue the research on understanding the structure and dynamics of complex networks, currently being undertaken by the candidate, but now directing the focus of analysis to information networks. For this, we propose the development and use of a framework to investigate dynamics on such structures by integrating data mining and machine learning approaches with network analysis. This may includes new visualization techniques, the use of natural language processing and new network-based approaches, such as mapping network dynamics to a feature space and modeling their evolution in terms of communities. Finally, we plan to illustrate the proposed framework through many applications, such as the analysis of social media data and of the structure and dynamics of science itself.
News published in Agência FAPESP Newsletter about the scholarship: