February 27, 2018 by Andreas Wierse, Geschäftsführer Sicos BW
Matthias Keller of Echobot reported his the experience with Smart Data Analytics and the SDSC-BW.
The Smart Data Solution Center Baden-Württemberg (SDSC-BW) and Matthias Keller from Echobot were able to secure a slot in the bulging lecture program of the Big Data Summit on February 28 in Hanau. The event coincided with the AI Summit, which focused on artificial intelligence and attracted over 1,000 people.
Matthias Keller reported on the experiences with Smart Data Analytics and SDSC-BW and also explained how the collaboration helped to identify new methods.
Every day, Echobot Media Technologies GmbH analyzes millions of texts from websites, social media and news sites for its customers. The analyzes help, for example, with marketing, sales or the optimization of press relations. Today, the company is a business information intelligence platform that analyzes billions of digital content and provides its customers with relevant real-time data and information. Companies that specialize in big data and smart data projects, such as Echobot, are constantly faced with the challenge of selecting and using suitable methods and algorithms in their analyzes. The selection and optimum configuration of these tools is crucial and has a lasting effect on the quality of the later results. The aim of the project with the SDSC-BW was to identify both hardware and software optimization potentials and to derive recommendations for the analytical methods used. The particular challenge in the project was the large number of different text classes or categories and the resulting large number of training data (“example texts”). The high demands required special hardware that Echobot itself was not available. Here, the SDIL could play its strengths.
At the event, we also took the opportunity to introduce the Smart Data Solution Center Baden-Württemberg. Success stories from practice were in the foreground, such as the potential analysis of Bilcare, Dr. Ing. Hartmann, Fuchs, Hermle, Herrenknecht and the Huber publishing house. In addition to many interesting discussions with other participants as well as exhibitors and sponsors, it a new potential analysis is already being prepared. The participation was completely successful!
Big Data All Hands Meeting @ KIT
October 12, 2017 by Parinaz Ameri
After two exciting days, the 2nd Big Data All Hands Meeting that was held in conjunction with the Smart Data Innovation Conference is over.
After two exciting days, the 2nd Big Data All Hands Meeting that was held in conjunction with the Smart Data Innovation Conference is over. 25 exciting talks from all over Germany demonstrated that Big Data is a major topic far beyond KIT and academia, as KIT’s vice president Oliver Kraft stressed when opening the conference.
The achievements during another year in German Big Data Research were presented. Different Big Data Projects reported on their successes as well as the known and newly rising challenges . The talks about legal and privacy issues at the beginning made it clear that Big Data will remain an interesting topic to practice from various aspects. The economical relevance became particularly obvious in the Data Innovation Community Meeting Industrie4.0. In addition, the societal relevance of the topic could be experienced looking at Big Data art at the ZKM in the evening.
The interdisciplinary aspect was a particular part of many scientific talks on applications, analytics and platforms. You could feel the enthusiasm of both researchers and industrial representatives during the whole conference. Chances for networking were well used. The hands-on-tutorial on the SDIL platform focused on the hot topics of data analytics and machine learning.
After a great meeting last year at SCADS in Dresden and a successful event this year, we are already looking forward to meet everyone next year at BBDC in Berlin.
Das SDIL arbeitet an der nächsten Generation kollaborativer Datenanalysewerkzeuge
July 30, 2017 by Michael Beigl
Erfahrungen in der Praxis zeigen, dass Datenanalysten einen großteil Ihrer Zeit während eines Analyseprojektes aufwänden für die Vorverarbeitung von Daten und die Erarbeitung eines zuverlässigen Modells zur Addressierung eines Analyseziels, beispielsweise der Vorsage eines Trends auf Basis vergangenen Verhaltens. Datenanalysten gehen dabei typischerweise iterativ vor: gegeben dem ersten Problemverständnis wird eine erste Datenvorverarbeitung und erste Modellbildung erzeugt. In Folgeschritten, wenn das Problemverständnis konkretisiert wird oder zusätzliche Fragestellungen in das Analyseziel aufgenommen werden, werden Vorverarbeitung und Modellierung angepasst.
Dabei werden die individuellen Iterationen typischerweise von einem bis weniger Datenanalysten gleichzeitig durchgeführt und diese Teams können über die Iterationen hinweg variieren. Das zentrale Problem in diesem Vorgehen ist, dass Veränderungen am Analyeprogramm häufig nur auf den ersten Blick das gewünschte Ziel erreichen, sondern stattdessen noch unbeabsichtigte Nebeneffekte erzeugen. Beispielsweise könnte ein beabsichtigter Filter von leeren Feldern in genau einer abgezielten Tabellenspalte, versehentlich alle Spalten einer Tabelle betreffen und dadurch auch Statistiken betreffen, die der Datenanalyst aktuell nicht im Blick hat. Um Fehler in Datenanalysen zu verringern und die Produktivität von Datenanalysten zu steigern, schlagen wir ein Test- und Validierungswerkzeug für von Analysten programmierte Analyseprozesse, vollständig computer-implementierte Datenanalysen, vor.
Unsere Arbeit umfasst die Spezifikation verschiedener Erweiterungen der populären Jupyter Notebook-Umgebung. Weitere Details finden Sie hier.
Projektvorstellung: Zustandsüberwachung von Dichtungslösungen von Trelleborg
April 27, 2017 by Johannes Kunze von Bischhoffshausen, Manager Digital Transformation and Internet of Things at Trelleborg Sealing Solutions
SDIL is a great initiative that is easing the journey for component developers and producers to fulfill the objectives of Industry 4.0.
Intelligent sealing systems have long been a dream of engineers at Trelleborg Sealing Systems. With the support of SDIL and in collaboration with our partners on the ‘Advanced Condition Monitoring for Sealing Solutions’ project, we hope that in some way we can make that dream a reality.
Industry 4.0 and the internet of things are becoming an important focus for manufacturers worldwide and there is a call for products to become smarter and more intelligent whatever and wherever they may feature. More and more users want intelligent products. SDIL is a great initiative that helps component developers and producers meet their goals. The engineers at Trelleborg have long dreamed of intelligent sealing systems. Together with SDIL we want to reach this dream.
Trelleborg Sealing Solutions is continuously working on the measuring and predicting of the condition of seals and it therefore runs a wide variety of instrumented tests on its test rigs. Measurements such as temperatures, speed, pressures and vibration are captured in very high frequencies. By leveraging Big Data Technology from its SDIL partners, Karlsruhe Institute of Technology (KIT) and IBM, Trelleborg will apply advanced machine learning in order to gain new insights, reduce testing costs and lay the foundation for advanced condition monitoring of sealing solutions in the field.
Huawei provides the SDIL with new, powerful hardware and software solutions
February 19, 2017 by Götz Brasche, CTO IT R&D and Director Central Software Institute
Huawei is pleased to be the new core partner to support the SDIL!
From now on, we provide researchers as well as the industry with a powerful instance of our FusionInsight hardware and software systems for big data analysis. The support for the SDIL, as one of the three major German competence centers for Big Data, is an important concern for Huawei. Besides the support for the research communities with a lot of know-how and research questions, Huawei provides core technology with FusionInsight Big Data. The computers were delivered to the SCC in early February and are currently being integrated as an important part of the SDIL platform. FusionInsight provides a comprehensive Big Data software platform for batch and real-time analysis using Hadoop and Spark technologies. The system leverages HDFS, HBase, MapReduce and YARN/Zookeeper for Hadoop clustering, along with Apache Spark for faster real-time analysis and interactive queries. Solr adds powerful full-text search in RTF documents (Word and PDF files), and Rich-APIs and development tools let you customize the system for a specialized data analysis. We are particularly pleased that SDIL (now also with the support of Huawei) has created a unique software and hardware platform that has just been recognized by the Big Data Value Association (BDVA) as one of only five Europe-wide innovation spaces. The offer from SDIL, which is specifically aimed at industrial data, is unique in Germany and lives from the cooperation between companies and research partners beyond SDIL core partners. We hope that based on the Huawei FusionInsight, further possibilities can be created to analyze data sets and streams with professional tools and to test new Big Data architectures (for example, by new streaming applications or by the integration of SAP HANA Vora with our Spark and Hadoop services). In this way, unique research infrastructure is available within the SDIL, especially for Big Data applied-research with real world data.