The Artifical Inteligence Group (LS VIII) is looking for another research assistant. Further details on the requirements, terms of reference and application can be found in the job description. (more...  )

For the COBRA project there is a vacancy of a research assistant. The project examines linguistic research questions and should now be supported by methods of data mining. Further details on the requirements, terms of reference and application can be found in the job description. (more...  )

For the EU project "Vista-TV", which deals with the analysis of Internet Television and the derivation of recommendations for it, LS VII seeks two research workers to 06/01/2012. Requirements, terms of reference and application details can be found in the job description. (more...  )

Rapid-I is based in Dortmund, Germany, and has been working on RapidMiner, a Data-Mining-Software, since 2001. With its wide range of other tools such as RapidAnalytics, RapidLab, RapidNet and RapidSentilyzer it could win over such clients as Siemens, Allianz and Pepsico. The website JTonEDM.com introduces Rapid-I and its software RapidMiner in a short overview. (more...  )

In heutigen Industrieanlagen zeichnen Sensoren während des Produktionsprozesses große Mengen an Daten auf. Aus diesen Daten wird schon während des laufenden Prozesses auf die Qualität des Endprodukts geschlossen. Produktionsbedingt kommt es während des laufenden Prozesses zu Veränderungen von Anlagenkomponenten und Messtechnik, die nur zyklisch instand gehalten werden können. Die Veränderungen spiegeln sich auch in den Prognosemodellen wieder. Es kommt zum Concept Drift. SMS Siemag und ein führender Hersteller von Grobblechen stellen für diese Arbeit aktuelle Produktionsdaten zur Verfügung. Im Rahmen der Bachelor, Diplom- oder Masterarbeit sollen Strategien zur Identifizierung von Concept Drifts und zur Stabilisierung der Prognosegüte entwickelt werden. Eine besondere Herausforderung stellt es dar, dass die Concept Drift Erkennung und Bereinigung in Realzeit geschehen soll. Der Schwerpunkt der Arbeit liegt daher auf der Auswahl, der Implementierung und dem Vergleich besonders effizienter Verfahren zur Entdeckung von Concept Drifts. (more...  )

In heutigen Industrieanlagen zeichnen Sensoren während des Produktionsprozesses große Mengen an Daten auf. Aus diesen Daten wird schon während des laufenden Prozesses auf die Qualität des Endprodukts geschlossen. Bisher wird von der Optimierung meist nur eine Zielgröße behandelt. Die Qualität des Endproduktes hängt aber oft von mehreren Zielgrößen ab, die sich obendrein widersprechen können. Dies kann nun als multikriterielles Optimierungsproblem formalisiert werden. Insbesondere muss eine geeignete Fitnessfunktion bestimmt werden. Dann können die Anwender aus den pareto-optimalen Lösungen Handlungsempfehlungen ableiten. Am LS8 stehen Sensordaten über den Produktionsprozess eines führenden Herstellers von Grobblechen zur Verfügung. An diesem Beispiel kann die Formalisierung von widersprüchlichen Zielgrößen als multikriterielle Optimierung untersucht werden. Dabei können Implementierungen in Rapid-Miner genutzt werden. Die genaue Aufgabenstellung wird darauf angepasst, ob es eine Bachelor, Diplom- oder Masterarbeit wird. (more...  )

Together with Kanishka Bhaduri and Hillol Kargupta, Katharina Morik has edited a special issue of the international journal Data Mining and Knowledge Discovery. The special issue on Data Mining for Sustainability including a comprehensive introduction is now online at http://www.springerlink.com/. (more...  )

Die Projektgruppe "Kooperatives Datamining mit vernetzen Robotern" wird am 22.12.2011 um 14:00 Uhr (s.t.) in den neuen Räumlichkeiten des Lehrstuhls 8 (Joseph-von-Frauenhofer Straße 23 in Raum 1.48) präsentiert.

Die Suche und Buchung von Hotels über das Internet wird heute üblicherweise über spezielle Portale abgewickelt. Die reine Filterung anhand von Suchkriterien führt häufig zur Ausgabe einer noch immer unüberschaubaren Anzahl von Hotels. Für die langfristige Bindung von Kunden an ein Portal ist es jedoch entscheidend, so schnell wie möglich Hotels anbieten zu können, die für die jeweilige Person (oder Personengruppe) tatsächlich geeignet sind. Mittels Methoden des Data Minings und maschinellen Lernens sollen Benutzerpräferenzen gelernt werden, die personalisierte und damit geeignetere Empfehlungen von Hotels ermöglichen. Hierzu werden vom weltweit führenden Portalbetreiber "Hotel Reservation Service" (HRS) Daten über Hotels, Portalbesucher, Kunden, Buchungen und Hotelbewertungen zur Verfügung gestellt. (more...  )

The annual ACM SIGKDD conference is the premier international forum for data mining researchers and practitioners from academia, industry, and government to share their ideas, research results and experiences. KDD-2011 will feature keynote presentations, oral paper presentations, poster sessions, workshops, tutorials, panels, exhibits, demonstrations, and the KDD Cup competition. KDD-2011 will run from August 21-24 in San Diego, CA and will feature hundreds of practitioners and academic data miners converging on the one location. (more...  )

The offer is for computer science students who are interested in machine learning and the analysis of big amounts of data. Start: immediately, 10 hours per week. For further information have a look at the job advertisement (in german only):

• Job advertisement

Being in the editorial boards of Knowledge and Information Systems (KAIS) and of Data Mining and Knowledge Discovery (DMKD), Katharina Morik happily presents the impact factors (2010) of some leading database and data mining journals:

• ACM Transactions on Information Systems (TOIS): 1.085
• ACM Transactions on Database Systems (TODS): 1.216
• Data Mining and Knowledge Discovery (DMKD): 1.238
• Information Systems (IS): 1.595
• Data and Knowledge Engineering (DKE): 1.717
• IEEE Transactions on Knowledge and Data Engineering (TKDE): 1.847
• Machine Learning (ML): 1.956
• Knowledge and Information Systems (KAIS): 2

Download the complete list

Neben YouTube und Co. wird das Internet mit zunehmender Bandbreite auch für klassisische Fernsehübertragungen immer interessanter. War IP-TV bisher meist für große Sportereignisse im Fokus, bieten Firmen wie z.B. zattoo.com bereits die Möglichkeit sich einer Vielzahl unterschiedlicher Kanäle zu bedienen, Sendungen online aufzuzeichnen und zu Archivieren. Aber wie findet man interessante Sendungen? Welche Informationen geben Aufschluß über Programme die mir gefallen? Lassen sich Spartensender allein anhand der Informationen aus den Video-Daten unterscheiden? In dieser Master-Arbeit geht es um die Extraktion von Merkmalen, die für die Klassifikation oder die Gruppierung von Sendungen, Sendern oder Fernsehzuschauern wichtig sind. (more...  )

The Feature Selection Extension für RapidMiner 5 contains some operators for feature selection and -weighting and for classification. All operators are also highly suitable for high-dimensional data, e.g. microarray data. New in this version are:

• RCCW - Recursive Conditional Correlation Weighting a very fast feature subset selection method.
• FCBF - Fast Correlation Based Feature Selection
• PAM - Classification by Shrunken Centroids
• BAHSIC - Backward Feature Selection via Hilbert-Schmidt information criterion
• t-Test - Computes a p-Value for the difference of the mean values between two classes
• Test Significance - Assumes normal distribution, then checks for equal class variances via F-test and afterwards computes p-Value via t-Test or Welch-test
• Benjamini-Hochberg-Correction - Performs the correction for FDR on significance values in an AttributeWeights object

Already available since older version are - amongst others - Recursive Feature Elimination (RFE) and minimum Redundancy Maximum Relevance Feature Selection (MRMR) / Correlation based Feature Selection (CFS) and a meta-operator for ensemble feature selection. The most recent version is available for free from SourceForge: https://sourceforge.net/projects/rm-featselext/ . (more...  )

RapidMiner is again the most popular data mining tool in KDnuggets poll. (more...  )

Graphics processor (GPU) architectures have evolved rapidly in recent years with increasing performance demanded by 3D graphics applications such as games. However, challenges exist in integrating complex GPUs into mobile devices because of power and energy constraints, motivating the need for energy efficiency in GPUs. While a significant amount of power optimisation research effort has concentrated on the CPU system, GPU power efficiency is a relatively new and important area because the power consumed by GPUs is similar in magnitude to CPU power. Power and energy efficiency can be introduced into GPUs at many different levels: (i) Hardware component level - queue structures, caches, filter arithmetic units, interconnection networks, processor cores, etc., can be optimised for power. (ii) Algorithm level - the deep and complex graphics processing computation pipeline can be modified to be energy aware. Shader programs written by the user can be transformed to be energy aware. (iii) System level - co-ordination at the level of task allocation, voltage and frequency scaling, etc., requires knowledge and control of several different GPU system components. (more...  )

An increasing amount of information is being encoded via ontologies and knowledge representation languages of some sort. Some of these knowledge bases are encoded manually, while others are generated automatically by information extraction techniques. In order to protect the confidentiality of this information, a natural choice consists in encoding policies with the same language as the ontology language. This approach led to so-called "semantic web policies". The semantic web is founded on two knowledge representation languages: description logics and logic programs. In this talk we compare their expressive power as *policy* representation languages, and show that logic programming approaches are currently more mature than description logics, although this picture may change in the near future. (more...  )

Emerging and envisioned applications within domains such as indoor navigation, fire-fighting, and precision agriculture still pose challenges for existing positioning solutions to operate accurately, reliably, and robustly in a variety of environments and conditions and under various application-specific constraints. This talk will first give a brief overview of efforts made in a Danish project to address challenges as mentioned above, and will subsequently focus on addressing the energy constraints imposed by Location-based Services (LBS), running on mobile user devices such as smartphones. A variety of LBS, including services for navigation, location-based search, social networking, games, and health and sports trackers, demand the positioning and trajectory tracking of smartphones. To be useful, such tracking has to be energy-efficient to avoid having a major impact on the battery life of the mobile device, since the battery capacity in modern smartphones is a scarce resource, and is not increasing at the same pace as new power-demanding features, including various positioning sensors, are added to such devices. We present novel on-device sensor management and trajectory updating strategies which intelligently determine when to sample different on-device positioning sensors (accelerometer, compass and GPS) and when data should be sent to a remote server and to which extent to simplify it beforehand in order to save communication costs. The resulting system is provided as uniform framework for both position and trajectory tracking and is configurable with regards to accuracy requirements. The effectiveness of our approach and the energy savings achievable are demonstrated both by emulation experiments using real-world data and by real-world deployments. (more...  )

• Program

(more...  )

Stefan Manegold from CWI Amsterdam will be giving a talk on the column-store DBMS MonetDB on 2011/02/11 um 16.00 at Room E23, Otto-Hahn-Straße 14.

Abstract:
Column-store database management systems have recently experienced a considerable popularity-boost. The underlying ideas, however, date back to (at least) the mid 1980's and the technology has been pioneered since the early 1990's in the MonetDB system, a column-store research prototype that has been developed into a complete SQL- and XML/XQuery-compliant column-store DBMS freely available in open source. Next to its column-store back-bone, MonetDB focuses on high-performance hardware-conscious algorithms, novel workload-adaptive query processing techniques such as "cracking", "recycling" and run-time query optimization, and extensibility at all layers of its software stack.

In this talk, we will provide detailed insight into MonetDB's column-store architecture and query-processing technology as available in open-source, discussing its benefits for data mining, OLAP, BI, as well as science workloads.

The new Collaborative Research Center SFB 876 "Providing Information by Resource-Constrained Data Analysis" starts the new year with a kick-off colloquium. The colloquium takes place on January 20th 2011 starting at 4 pm at auditorium E23, Otto-Hahn-Straße 14, TU Dortmund University campus. For further information about the program and speeches please have a look at the attachment.

• Schedule
• Monitoring the Data Tsunami (Johannes Gehrke) , Download slides
• Model predictive control and self-learning of thermal models for thermal control for multi-core platforms (Luca Benini) , Download slides

First presentations and pictures available on the MODAP workshop website. (more...  )