Prof. Dr. Emmanuel Müller

Chair for Knowledge Discovery and Data Mining

Knowledge Discovery and Data Mining, as part of many scientific and industrial applications, does not end with the execution of algorithms. With data mining algorithms, resulting in discovery of unknown, novel, and unexpected patterns, one should aim at assisting humans in their daily decision making. On the one side, we investigate efficient algorithms, which scale with size and complexity of the data. And on the other side, our algorithms generate verifiable knowledge for human users.

Algorithms for Big Data Analytics 
Our research addresses theoretic challenges in correlation analysis, (un-)supervised feature selection, cluster and outlier detection as well as practical challenges in efficient computation of these models in large and complex data. The development of novel techniques for heterogeneous data spaces is a particular challenge in this area. We overcome the information loss of traditional techniques on homogeneous data sources and utilize the huge potential that is still unused in heterogeneous databases. Our group investigates algorithms for the selection of relevant attributes in high dimensional data, correlations in multivariate data streams, and homophile structures in attributed graphs.

Verifiable Knowledge for Human Users
Our research aims at an easy to understand presentation of data analytics results. We represent intrinsic dependencies between different information sources for human users. This includes exploring the automatic extraction of dependencies and pattern descriptions. This is an important research contribution for many applications where patterns have to be verified by the users. Human users require such descriptions of potential reasons for each of the detected patterns. This includes rule-based descriptions for unexpected patterns, semi-automated data exploration, and schema extraction.

Data Mining at Hasso-Plattner-Institute