Prices, Awards, and Achievements

This page provides information about the standing of database group of Hans-Peter Kriegel and the impact of its work.

Current Projects

Cooperations

Theseus

The vision of the application scenario is to establish syntax and semantics based on image vocabularies to build an intelligent scalable robust search engine for the medical domain, with applications in Clinical Decision Support and Computer Aided Diagnosis.

Sendsor

Controlling convalescence of a patient is not an easy task if the convalescence has to be be measured in terms of activities. In order to solve this problem, we are developing a system that gathers, joins and analyses the data of multiple miniaturised activity monitors, which are mounted to the patient whose convalescence should be controlled.

Additional to the processing of raw data, the prototype also involves the visualization of the results in a suitable manner for attending physicians.

Similarity Search

Adaptive Similarity Search

An important aspect when looking for similar data objects is the underlying concept of similarity. Since this concept might depend on the current user or even the current situation it is important that a similarity search system is capable to adapt to multiple similarity functions. Further problems in adpative similarity search include capturing the user intention, learning similarity functions and indexing objects for similarity queries with varying concepts of similarity.

Similarity Search in Spatial and Multimedia Databases

Ähnlichkeitssuche für Medizin, Biologie und Multimedia Adaptable Similarity Models, Quadratic Form Distance Functions, Ellipsoid Queries (exact and approximate), Reduction of Dimensionality, Optimal Multi-Step k-Nearest Neighbor Search, k-NN Classification

Efficient Reverse Nearest Neighbor Search

This project explores data structures and algorithms for various variants of the reverse k-nearest neighbor (RkNN) problem, including e.g. classical RkNN queries, RkNN rankings, constrainted RkNN queries, etc.

Data Mining

Data Mining and Routing in Traffic Networks

Modern spatial databases describing traffic networks provide a variety of information about the connections of two locations. For example, a database might store the distance, the speed limit, the altitude difference or the number of traffic lights for each road segment. Thus, a driver might want to consider various criteria at the same time.

However, employing ordinary shortest path routing would require to select a single criteria or define a preference function by weighting each criterion. For example, a user might enter that his major preference is driving the fastest path with a weight of 80%, but still wants to consider driving distance with a weight of 20% due to fuel costs. By considering travel distance as well as travel time, the selected route might be considerable shorter (and thus cheaper) and only slightly slower than the fastest path.

Routing with multiple means of transportation

Clustering High Dimensional Data

High-dimensional data is prevalent in many applications and poses several new challenges for clustering algorithms. In this project, we explore the fundamental problems occurring in high dimensional spaces (a.k.a. "curse of dimensionality") in light of clustering and develop new, specialized methods for efficient and effective cluster analysis in high dimensional data.

ELKI: Environment for deveLoping KDD-applications supported by Index-structures

The software system ELKI presents a large collection of data mining algorithms and support of database queries by arbitrary index structures. ELKI also enables to work on arbitrary data types given supporting data classes and distance functions.

Outlier Detection

Data Mining in Bioinformatics

The research group of Hans-Peter Kriegel has a long standing tradition in contributing database support and data mining methods to the application domain of bioinformatics. Among these methods are classification and similarity search in 3D molecular databases, database support for the one-to-many protein docking search, data mining solutions to prediction of protein function, protein structure, and protein subcellular location.

Knowledge Discovery in Large Collection of Complex Objects

High Performance Data Mining

This project deals with performance issues of data mining algorithms. In particular, we explore data structures and new algorithmic concepts to make data mining solutions scalable to very large databases.

Former Projects

• Effiziente Anfragebearbeitung in Geo-Datenbanksystemen
• Geometrie-orientierte Ähnlichteilsuche in Geo-Datenbanksystemen
• Interoperablilität heterogener Informationsressourcen
• Visuelle Unterstützung für Anfragen an große Datenbanken

Knowledge Discovery in Databases, Data Mining

Clustering and Outlier Detection (DBSCAN, OPTICS, OPTICS-OF), Classification, Database Primitives, Spatial Characterization, Spatial Trend Detection, Visual Data Mining

Spatial Data Management in CAD Databases

Approximation of Surfaces and Solids, Spatial Query Processing, Spatial Access Methods, Collision Detection and Within-Distance Queries for Digital Mockup, Similarity Search, Engineering Data Management

Access Methods for High-Dimensional Data Spaces

High-Dimensional Indexing, Parallel Index Structures, Relational Implementation of Spatial Access Methods