Summer Term 2012

We present different topics of our research every Thursday at 12:00 noon in room 305, Obermarkt 17. Please feel free to contact us!

02.08.2012 Abhishek Awasthi "Clustering Algorithms for detecting Money Laundering using Graph Theory and Metrics of Social Network Analysis"

contact: aawasthi(at)hszg.de

We present results of the Master thesis "Clustering Algorithms for detecting Money Laundering using Graph Theory and Metrics of Social Network Analysis". Useful aspects of graph theory and SNA metrics', Page Rank and Hub & Authority algorithms including some standard clustering algorithms are discussed. Also an improvement of the k-means clustering algorithm is suggested. Further, we elaborate on a Genetic Algorithm (GA) approach for community detection in social networks.

26.07.2012 André Mattusch "Web-based Business Process Modeling"

contact: mattusch(at)iqnow.de

We introduce the idea of a web-based business process management system in an early stage. The talk describes, which features we are going to realize, including the modeling of business processes, quality management, surveillance of business processes as well as process documentation and evaluation. This is the basis for process improvement and optimization, which we also plan to support. The application is planned to fit particularly well for Small or Medium Sized Enterprises (SME).

14.06.2012 Robert Kloß "Energy Efficiency Benchmarking for SME"

contact: s3roklos(at)stud.hszg.de

The energy efficiency is one of the most important parameters to measure the overall effectiveness of a company. To compare the energy efficiency, a benchmark has been developed which can now be generated by a fee-based web application. For this a large number of corporate data and data of the industry are compared. The aim is to show the company where it stands in comparison to the rest of the companies in the industry.

07.06.2012 Markus Ullrich "Job Manager for Data Mining Tasks in the Amazon Cloud"

contact: m.ullrich(at)hszg.de

The thesis is about the design, implementation and test of a job manager to transparently execute datamining tasks at the amazon cloud. More exactly it is about an extension to an existing desktop tool called PdM-Toolbox which is used to identify deterioration pattern for machine parts by data mining observations and maintenance activities. The goal is to let the user choose how much he wants to pay for certain datamining tasks that will automatically be executed at the cloud. That includes the transfer of jobs into the cloud, the creation of instances, the job execution at the cloud, the shutdown of instances after completion and the transfer of results back to the client.

31.05.2012 Nico Dittmann "Comparison of Open Source ERP-Systems"

contact: n.dittmann(at)hszg.de

We present an overview of open source Enterprise-Resource-Planning (ERP) systems. They are compared on the basis of different characteristics like support, community, features and many more and how they are suitable for further development and the integration of new functionalities. Furthermore, the idea of my masther‘s thesis is presented. In detail, we want to combine AMOPA, an image processing framework, with ERP systems for quality purposes. In addition, a system based on the Compute Unified Device Architecture (CUDA) will be implemented so we will be able to use the power of GPUs for processing high performance algorithms.

24.05.2012 Waheed Ghumman "Cloud Computing and Grid Computing 360-Degree Compared"

contact: w.ghumman(at)hszg.de

Cloud computing has become another buzzword after Web 2.0. However, there are dozens of different definitions for cloud computing and there seems to be no consensus on what a cloud is. On the other hand, cloud computing is not a completely new concept; it has intricate connection to the relatively new but thirteen-year established grid computing paradigm, and other relevant technologies such as utility computing, cluster computing, and distributed systems in general. This paper strives to compare and contrast cloud computing with grid computing from various angles and give insights into the essential characteristics of both.

10.05.2012 Stefan Dahms "SQUIDD - A Framework for Distributed Data Mining on Large Data Sets"

contact: stefan.dahms(at)googlemail.com

In the area of e-commerce one of the key success factors for companies is personalization and its integration into customer-related processes. The data and data sources necessary are broadly available due to e.g. the increasing role of computer assisted process management, falling prices for computing power and storage and the proliferation of data warehouses. In order to use the data a variety of tools and frameworks are available. This development has been recently coined by the term "Big Data". Nevertheless the generation of possibly useful information out of large data sets is only the first step. In order for a company to create a measurable business value from the utilization of "Big Data" it is also necessary to reintegrate the results of the data mining process seamlessly and in real time back into its core business workflows. SQUIDD attempts to fills this gap by combining a framework for distributable, horizontally scalable machine learning algorithms with the provision of data mining functionality over web services tied together in a modular architecture.

03.05.2012 Markus Ullrich "Finding the Deterioration Needle in the Log Haystack"

contact: m.ullrich(at)hszg.de

The thesis is about the design, implementation and test of a job manager to transparently execute datamining tasks at the amazon cloud. More exactly it is about an extension to an existing desktop tool called PdM-Toolbox which is used to identify deterioration pattern for machine parts by data mining observations and maintenance activities. The goal is to let the user choose how much he wants to pay for certain datamining tasks that will automatically be executed at the cloud. That includes the transfer of jobs into the cloud, the creation of instances, the job execution at the cloud, the shutdown of instances after completion and the transfer of results back to the client.

25.04.2012 Nico Schlitter "DistributedDataMining Project"

contact: n.schlitter(at)hszg.de

distributedDataMining (dDM) is a scientific computing project that provides the computational power of internet-connected computers to its scientific partners in order to perform research in the various fields of Simulation, Data Analysis and Machine Learning. The project uses the Berkeley Open Infrastructure for Network Computing (BOINC), which is an open source framework for the distribution of research related tasks to a large number of participating computers. The dDM project became available to the public in March 2010.