Applications, for instance in the areas of artificial intelligence or information technology in general, often represent a system with a large data set at its core. This system operates on heterogeneous, possibly unstructured data (texts, diagrams, pictures, time-series data, video data). Algorithms process these data to fulfil an implicitly or explicitly stated purpose in the application, e.g., learning for obtaining or improving a model of how data might have been generated. The model captures application-related aspects of data semantics that humans would perceive when interpreting data. However, data processing goes beyond data handling and often extends to query answering. Algorithms for answering queries on models, also called inference algorithms, allow computational systems as well to further interpret data in a semantically meaningful way. In particular, if systems make decisions or act in an even more general sense, systems need to provide explanations for its actions or, more generally stated, act comprehensibly in the given environment. Dealing with these complex requirements in such systems make up modern aspects of data processing as well as data science.
Specifically, we are, among other things, interested in:
- Statistical relational AI, specifically, probabilistic inference (probability- as well as decision-theoretic) in relational domains
- Text understanding, specifically, annotations and context representations
- Human-aware AI, specifically, decision making in human-aware agents