Advances in data analytics and machine learning have provided new ways to detect and reason about patterns in large data sets, to extract knowledge from vast corpora of unstructured information, and to support decision-making through predictive modeling and analytics. Despite these advances, the application of such Big Data Science techniques in practice is still a challenge. Apart from their large size, real-world data often exhibit complex characteristics - such as relational dependencies, time stamps and temporal correlations, noise and uncertainty, or a combination thereof - that hinder a straight-forward application of standard algorithms.
Addressing this challenge, our goal is to provide scalable data analytics and machine learning techniques for data with complex relational and temporal characteristics. Developing efficient algorithms for (i) graph mining and network analytics in time series data, and (ii) pattern recognition in uncertain relational data, our research is centered around two current themes in data science and statistical relational learning.
Our interdisciplinary approach integrates methods from computer science, applied statistics and (random) graph theory with modeling and simulation techniques from statistical physics. We use our methods to answer interdisciplinary questions in the area of collaborative software engineering and computational social Science.
You can find more details on our different lines of research in the menu on the left.