Termine Forschungskolloquium DIMA

Abstract:

"Folksonomy based data standardization"

Feed Mashups integrate different Web 2.0 data sources. In this area one has to deal with common data integration problems such as in ETL and EII. The problem with web based data integration is that data sources often lack in structure. Furthermore entities in data sources have different representations for the same domain.
IBM MashupHub is a situational application environment helping “business users” to create Mashups. Those users often have problems to integrate their data because they don’t know the common touch points of their data sources.
The solution presented in this talk makes use of folksonomy in enterprises. We claim that there are some expert users who might know how to join domains and there are different standardization services in the web or the company’s intranet.
The system we discuss is able to analyze Mashups created by those users and store the information in a repository. Furthermore it is able to give suggestions on how to integrate unknown data sources.

Abstract:

THESEUS ist ein Programm und ein sogenanntes Leuchtturmprojekt der Bundesregierung. An THESEUS sind mehr als 30 Partner beteiligt. Die Präsentation des Projektes wird Ausgangspunkt für die Beschreibung der Aufgaben der sogenannten Begleitforschung sein. Dies wird kein primär technischer Vortrag, sondern - so habe ich den Wunsch von Herrn Kollegen Markl verstanden - ein "förderpolitischer" Vortrag. Technische Fragen werden selbstverständlich, wo immer nötig, eine Rolle spielen.

Short Abstract:
Das ModelBus Repository ist ein Model Repository, welches im Kontext
von Förderprojekten von Fraunhofer FOKUS entwickelt wird. Es ist
zentraler Bestandteil der ModelBus Architektur, welche die verteilte
modellgetriebene Entwicklung unterstützen soll. Vorgestellt werden die
Architektur, Funktionalität und Zukunft des Repositories und Eclipse
Team Providers.


Short Bio:
Cristian Hein und Michael Wagner sind die beiden Hauptentwickler des
ModelBus. Beide haben ihr Diplom an der TU-Berlin gemacht, und arbeiten
seit mehreren Jahren für Fraunhofer FOKUS als Forscher im Bereich der
modellgetriebenen Softwareentwicklung und Automatisierung.

Abstract:

The paradigm of model-driven system design rapidly dominates software
system engineering. The core element of all model-based technologies
is model transformation, which serves for mapping an engineering model
into mathematical representation for analysis or into a run-time code.
Despite of the fact that transformers play a similarly central role in
the efficiency, quality and practical usability of MDA, as compilers
do it in traditional software engineering, transformation based model
engineering is still in its very initial and premature state.

The talk summarizes the main conclusions gained in a series of European
projects at the Budapest University of Technology and Economics and
Optxware Ltd. in different fields of application of model transformations,
including tool and data integration, automated generation of safety-critical
system architectures, derivation of validation and validation models and
workflows for certification. In particular, it addresses the aspects of
(1) traceability, (2) rapid transformation development, (3) scalability
approaches for large models, (4) adaptivity of transformations with a
special emphasis to domain specific languages, (5) model and transformation
complexity and quality assessment.


Bio:

Professor András Pataricza received his diploma in Electrical Engineering
in 1977 from the Technical University Budapest, his PhD and Doctor of the
HungarianAcademy of Sciences degree in 1988 and 2008, respectively.
Since 1977 he is with the Department of Measurement and Information Systems.
He is the leader of the Fault-Tolerant Systems Research Group.
He was a visiting professor at the University of Erlangen-Nuremberg
between 1993 and 1994, and in 2003.
He is the author, co-author or editor of 13 books, 7 book chapters,
25 journal articles and 110 conference papers.
He has been the Hungarian project leader of several EU projects (HIDE,
HIDENETS, RESIST, SENSORIA, GENESYS and DECOS) and many academic and
industrial research projects. He was PC member of DSN, EDCC, DDECS, ISAS,
FTCS etc. conference series and SC member, general co-chair and program
co-chair of EDCC, ISAS and DDECS conference series. He is founder and
president of a spin-off company named OptXware established together with
the members of his Research Group in order to promote model-based computing
with a special emphasis of the aspects and consolidation of dependability.

Abstract

Mahout ist ein Apache Projekt unter der Schirmherrschaft von Lucene.
Das Ziel von Mahout ist es, skalierbare, Apache lizensierte Machine
Learning Bibliotheken zur Verfügung zu stellen. Der Vortrag wird
einen Überblick über das Projekt geben: Nachdem kurz auf den
Hintergrund des Projektes eingegangen wurde, werden die bisher
verfügbaren Algorithmen knapp vorgestellt sowie auf zukünftige
Entwicklungen eingegangen.


Bio

Isabel Drost ist Organisator des Apache Hadoop Get Together in Berlin.
Sie ist Cogründer von und Committer bei Apache Mahout. Isabel arbeitet
für die neofonie GmbH Berlin. Nachdem sie ihr Diplom von der Hochschule
für Technik und Wirtschaft Mittweida erlangt hat, arbeitete sie als
wissenschaftlicher Mitarbeiter an der HU und war für sechs Monate bei
Google Zürich als Praktikant angestellt.

Abstract

Semantic processing of natural language is one of the oldest problems in
machine learning and still far from being solved. By now, low-level tasks
including part-of-speech tagging and named entity recognition are well
understood while complex tasks such as parsing, machine translation, and
sentiment prediction are still lively subjects of ongoing research. The
talk focuses on the identification of relations in sentences. Starting
from classical (pipelined) approaches we'll derive state-of-the-art
techniques by addressing complex tasks in a single optimization problem.
We'll also learn about two naturally arising problems: Firstly, the trade-
off between performance and execution time and secondly, the quest for
annotated data.


Bio

Since October 2007, Ulf is a postdoc in the Machine Learning Group at
Technische Universität Berlin. Prior to joining TU Berlin, he worked at
Max Planck Institute for Computer Science in Saarbrücken and at Humboldt-
Universität zu Berlin. Ulf received a Diploma in Computer Science in 2003
from Technische Universität Berlin and a Ph.D. (Dr. rer. nat.) in 2008
from Humboldt-Universität zu Berlin.