@prefix : ](http://denigma.de/resource/>)] .
@prefix dc: ](http://purl.org/dc/elements/1.1/>)] .
:Grant_Call a owl:[Class](/blog/619/) .
](http://www.bmbf.bund.de/foerderungen/21340.php>)] a Grant_Call ;
dc:title “Management und Analyse großer Datenmengen (Big [Data](/blog/824/))” ;
:deadline “12. Juli 2013” ;
:size “10 pages” .
Project description
Duration (up to 3 years)
Topic and Aim
Sketch of concrete aimed application scenarios
State of the Art in Science and Technique
Consideration of ongoing national and international research
Novel approaches
Description of the Proposed Solution
Benchmarks regarding size and Type of Data as well as aimed Improvements regarding processing efficiency and data integration
Plans & Prerequisites for Application of Results in Science (and maybe economics)
Working Plan and distribution of work among cooperating partners
Finance plan and amount of funds
Short description of applying organizations
Relevance of the research aims
Scientific-technique Quality of the Approach
Qualification of Partners
Project management and collaborating structure as well as cost-usage relation
Technological and eventually economic potential
Improved Performance via Efficiency and Integration
Effect on the German economy in regard to the contributions to different Big Data using scientific disciplines and contributions to interdisciplinary Research fields of data Data Science
Quality and Feasibility of the Realisation Plan, Sustainability of Results and Potential to broad effect in the application domain and beyond
Relevance to practice of research results and potential approaches for solution
Effect on the German Economy and Science in regard to different Big Data using application domains
We propose to use Semantic Web Technologies and Artificial Intelligence to address Big Data challenges especially in the Domain of Biomedicine with a Focus on Aging.
The Size of scientific and economic produced Data grows continuously and exponentially. The Management of giant Data Amounts is one of the most important Challenges of the post-industrial era. The Development of intelligent Technologies for the handling of Big Data for Science and Industry is of high value.
Crowdsourcing is means of performing Tasks which are difficult for Computers including Entity Resolution, Audio Transcription, Image Annotation, Sentiment Analysis and Document Summarization as well as Editing.
Besides the Fact that Humans are more accurate than Machines at some of these Tasks, using Humans for annotating large Datasets becomes infeasible on the long run if dealing with Web-scale data. Tens of millions of Information in form of Documents, Images, and Blog Posts are uploaded on a daily basis while labeling by a Human becomes ineffective.
The Solution is to integrate Machine Learning into Crowdsourcing Workflows to annotate very large Datasets faster, more efficient (i.e. cheaper) and even more accurately. Specifically given unlabeled Data and a Classifier (based on Ontologies) that, once trained sufficiently, can identify Concepts and annotate Data and learn from it and from the Crowd.
Biomedicine is both a big Producer and User of Big Data.
We propose a Combination of several cutting edge Technologies and use Biomedicine as a Case Study specifically the major biomedical Problem of [Aging](/blog/210/) which has a high relevance to the German Economy.
Crowdsourcing (perhaps with a Focus on distributed Computing and semi-automated Annotation) of Big Data Tasks in Biomedicine and Aging.
Ontology Management and more efficient Queries of Ontologies in the Context of Model Management and Semantic Systems Biology.
Biomedical Data is extremely interconnected, rendering it perfectly suitable for Graph Representations.
Biomedical Data
Artificial Intelligence
Decoding of unstructured Information
Encoding into structured Data
Distributed Graphs
Application of Artificial Intelligence
Setting up Crawlers that gather unstructured Data and convert it into structured Formats (following Web Standards) via Natural Language Processing and Semantic Web Technologies.
Automated Semantic Annotation of diverse Media and Data Resources including Texts, Images and Videos.
Validation of Knowledge and training of Machine Learning Algorithms via the Crowd continuously improves the automatic Conversion.
We will use NOSQL databases, in particular Graph Data Representations and try to scale them up (which is ongoing research).
Enabling Semantic Search and especially Exploratory Search of the constructed Knowledge Base.
Visualization of the data in four dimensions (4D) via Graphics Engines used in the Gaming Industry.
University of Rostock
Hasso-Plattner-Institute Potsdam
The sole provision of scientific Data from different Sources via Distributed Open Science can lead to shorter Search and processing times and more efficient ways for Discoveries.
The Integration and Summarization of this Data in the Field Research and Development can be achieved by a combination of specific cutting-edge Technologies
The Application of Semantic Web Concepts enables the Analysis Big Datasets as well as Combination of diverse heterogeneous Data Sources.
For this Database-oriented Approaches utilizing emerging NOSQL databases as well as Information Retrieval via Semantic Search will be applied. Approaches for dynamic streaming of Datasets will be used in a decentralized fashion.
The automated usage of Machine Learning Techniques on structured Data gathered via semi-automated means from heterogeneous as well as distributed and unstructured Data is an effective way of scaling with the growing amount of Data on the Web.
Semantic Queries hidden behind constrained Natural Language and 4D Visualization of the Data enables the real-time Analysis for the rational Usage by humans.
Real-time and interactive Simulation on the basis of big data will allow ad-hoc Decisions on the basis of generated Alternatives.
The processing of large-scale Ontology modelled Knowledge Bases will be efficient with the use of the Architecture and Algorithms of High-Performance Computers including Parallel Computing.
These described scalable Techniques and the User-friendly Presentation of huge Data Amounts makes the further Interpretation of Data by Humans possible.
Such novel 4D Visualization Techniques can greatly help ot manage getting insights in the increasing data amounts.
The developed Applications and Approaches based on the Combination of the above mentioned Technologies and Rationalities will not solely be restricted to the Biomedical Domain. A Prerequisite of the this Project is the Development of Techniques and Tools which are not restricted to a single Branch or scientific Discipline, but as much as possible will be utilizable for high-tech strategies of other fields including those covering social and economic problems.
During the project extensive collaboration with other projects that are within the funding priorities of this call will be established and maintained. Further public engagement will be used extensively as well as transfer knowledge and benefit the German People.
http://www.bmbf.bund.de/foerderungen/21340.php is similar to http://www.bmbf.de/foerderungen/21339.php
[is a](http://www.pt-it.pt-dlr.de/de/3095.php http://www.pt-it.pt-dlr.de/de/3095.php is a Conference ]
has title “Big Data: Datenschätze in Wissenschaft und Wirtschaft heben” ;
termin “3. Juni 2013, 10:00 Uhr - 16:45 Uhr”
place “Berlin”
TitanDB with Cassandra scales very well. It can be used for queries like "show me thing connected in some way". For math-like tasks libraries such as signal-collect or Giraph should be used. There are also next generation of graph processing software - Pregel, scales very well.
Giraph builds upon the graph-oriented nature of Pregel but additionally add fault-tolerance to the coordinator process with the use of ZooKeeper as its centralized coordination service.
Pregel is a next generation graph processing software that scales very well.
Apache Mahout is machine learning library with the goal to build scalable machine learning libraries [http://mahout.apache.org/],like clustering, classification, collaborative filtering and frequent pattern mining.
AI/ML cloud platform - We create a cloud system with basic API for performing general AI/ML tasks. For example, NLTK processes can be crowdsourced and distributed computation to many machines.
The same is true for image processing.
one size fits all solutions like cloud solution or high-computing service.