Winter Term 2015/16

We present different topics of our research every Tuesday at 11:30 am in room 103, Obermarkt 17. For news about the EAD-Lunch talks and seminars please feel free to subscribe to (Register here:!forum/ead-public)

04.02.2016 Jens Heider "Introduction to Kiwigrid"


Kiwigrid offers a smart grid management platform consisting of hard- and software. They provide tools to integrate and manage distributed energy generators, consumers, storage systems, e-mobility and the grid itself in microgrids.

12.01.2016 Daniel Müssig "An Applied Study of the Privacy Preserving Neural Network with Horizontally Partitioned Data"


This lunch talk we discuss an algorithm for privacy preserving data mining (PPDM) using a neural network with special emphasis on distributed calculations over horizontally partitioned data in the enterprise context. The focus of the talk will be on the implementation and performance issues of the distributed neural network algorithm over horizontally partitioned data, their practical applicability and evaluation. The algorithm is broken down into basic and specific operations for privacy preserving computation and distributed over several parties. The distributed and privacy preserving implementation is compared with a classical approach, showing that in general it is applicable and even performs better because of the parallel implementation.

05.01.2016 Waheed Ghumman "Automated Cloud SLA Life Cycle - State-of-the-art, Gaps and Proposed Solution"


Cloud computing has been established as a ubiquitous model for on-demand computing with regnant effects across IT infrastructures, software architecture, business, data storage, applications and services. Cloud services are most often bounded by an agreement between cloud service provider (CSP) and a cloud service user (CSU), termed as service level agreement (SLA), to define the quality of service (QoS) parameters. The life cycle of an SLA generally includes requirement specification, negotiation over QoS parameters between a CSU and a CSP, monitoring of SLAs to ensure QoS requirements and management of SLAs. Manual or semi-automated approaches for complete SLA life cycle and for individual stages diminish true utility of cloud features like cost efficiency, timeouts availability of cloud services and lesser administrative overheads. In this talk, a motivation for the automation of the complete SLA life cycle is discussed. A detailed overview of state-of-the-art approaches/models for SLA specification/standardization, automated negotiation, monitoring and management is also presented together with their existing problems, drawbacks and gaps. An approach for object oriented SLA structure, an automated negotiation model targeting specifically cloud services and initial results to illustrate the effectiveness of the proposed solution is also presented.

08.12.2015 Florian Haje "Clustering of Tf (-Idf)-Vectors for the NXTM Project"


Tf-Idf weight is a reliable way to compute how similar texts are with regards to content. This is the theoretical basis, and clustering of Tf (-Idf)-vectors is a practical way to find texts with similar topics. The NXTM-Project uses cluster structures to find documents containing keywords or being similar to a given document with regards to topic in an efficient way.

01.12.2015 Jens Leuschner "Algorithm to Optimize Products Selection Based on the Acquisition Costs for the VATI Project"


Ambient Assisting Living (AAL) programme promotes innovation and development for providing services and systems for the elderly people of the society to age well at home, in the community and at work, thus increasing the quality of their life. VATI is one such project on these guidelines. The aim of the project is to create an online portal, where elderly people can answer some questions to get some offers of products they can use for a dignified life at home. The questions and their resulting answers are transferred to some product classes, which contain different products with several dependencies to each other. This talk will present you an algorithm which selects a set of these products by optimizing the acquisition costs and their benefits to the customer.

24.11.2015 Franz Scheffler "Eine Kurzeinführung in verschiedene Modelle der Computerlinguistik"


Die automatisierte Gewinnung von Informationen aus Texten mit natürlicher Sprache ist das Arbeitsgebiet der Computerlinguistik. Dieser Vortrag soll eine Kurzeinführung in bestimmte Modelle darstellen, die heutzutage zur Klassifizierung bestimmter Textpassagen in beliebigen Dokumenten eingesetzt werden. Es werden außerdem einige Vor- und Nachteile dieser Modelle kurz dargelegt.

17.11.2015 Adam Bartusiak "Structured data search with SIREn"


SIREn is a open-source full-text search engine especially suited for nested and schemaless data. It can be used standalone, similar to Lucene, and it comes in the form of Elasticsearch and Solr plugins. SIREn has its own data architecture which is suited for nested documents such as JSON and XML. The talk will give a short overview of SIREn’s history, concepts, its indexing model and query language.

27.10.2015 Jens Heider "Climate Strategy for Municipalities to reduce their Energy Consumption and Greenhouse Gas Emissions"


The talk will present a set of four web based services which form the base of a climate strategy for municipalities to reduce their energy consumption and greenhouse emissions. Starting with a solution for telemetrical measuring and visualizing of consumptions in real time, these consumption values can be evaluated regarding to costs and emissions. One-Click-Reports are possible. A third services supports the complete workflow to create complex greenhouse gas balances. On the one hand it supports the required data input by the municipality. On the other hand it manages required statistics and background values to make the creation process for experts as easy as possible. A last service is intended to support workshops to create scenarios, which consists of concrete action to reduce greenhouse gas emissions in all relevant sectors.

13.10.2015 Markus Ullrich "SERVING Project: Introduction"


The research project SERVING, located on the 6th Energy Research Programme – Energy Saving and Energy Efficiency, has the aim to operate intelligently the distribution grid with the help of a Service-Platform. The infrastructure of distribution grid will be operated in an optimal way under consideration of decentralized Renewable Energy Plants and the use of controllable loads. Thereby, the highest flexibility potentials of consumers are available to energy suppliers. The Service-Platform is not only responsible for the communication, it also performs the state-estimation of the distribution grid and takes various requirements of the market into account. In case of network bottlenecks or quality problems the load is shifted by help of a new load-allocation algorithm. This method is practically tested on the basis of heat storage systems and water supply systems. The development of the Service-Platform-Distribution-Grid is a sustainable step within the evolution of the distribution grid into a smart grid and provides a holistic solution for reliable distribution grids in the future.

02.09.2015 Maximilian Kaiblinger "Energy saving potential assessment based on the eco-design guidelines of the European Union"


The European Union is eager to provide guidelines for consumption of energy. Hence, the EU developed directives deal with energy end-use efficiency, energy services or energy labeling. These directives provide a framework for companies on how to measure and label the energy consumption for their products. A variety of appliances, like TVs, boilers, air conditioners, etc. are legislated by the EU and offers an impulse for further investigation. Based on these directives a first "lighting assessment excel tool" is developed which helps to estimate the energy use over a certain lifetime.