University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Graduation Talks

title

Visual Analytics in Multi-Dimensional Databases

speaker

Matthias Schäfer, University Konstanz
Konstanz, Germany

date & place

Wednesday, 01.12.2010, 16:15 h
Room C 252

abstract

The amount of data being collected and stored by commercial organizations is increasing at a fast rate and hence implementing intelligent and flexible visual analytics systems is important in many business cases. The specified research project focuses on analyzing multi-dimensional event time series data and also multimedia data. An event is defined as a single, time-stamped item. Event time series data is a sequence of events which occurs 0 to N times with multiple values per time stamp. Multimedia data includes audio, image, 3D or video data. The common characteristic of these data types is that it is possible to calculate high dimensional content-based feature vectors that represent the multimedia data.
All these types of data are difficult to visualize. So the challenge is to analyze and visualize this data in such a way as to gain new insights and extract knowledge from the data. The goal is to achieve a close interaction between the algorithmic and the visual parts of the visual analytics systems and a high-flexibility so that the user can delve deep into the data to get details on demand and can also give feedback to steer new analysis tasks.
Applications in fraud detection, finance and multimedia projects will be used to demonstrate the flexibility of the systems in establishing a separation between the data and the visualization and also the ability to adapt to several domains. Visual analytics algorithms have been developed to overcome challenges such as the visualization of different alignment strategies to deal with temporal gaps, aggregation and change calculations of the data, pattern detection and automatic feature selection.