HOBBIT @ Graph-TA
1 like358 views
The HOBBIT project focuses on holistic benchmarking of big linked data, addressing key aspects of the linked data lifecycle, including performance evaluation of various technologies and methodologies. The project aims to establish a universal benchmarking platform and gather real performance indicators from the industry over a duration of three years. It involves multiple partners across Europe and seeks to improve efficiencies in data generation, analysis, storage, and visualization.
HOBBIT @ Graph-TA
- 1. Holistic Benchmarking of Big Linked Data HOBBIT Irini Fundulaki Institute of Computer Science - FORTH Greece 4th Graph-TA Barcelona, Spain, March 3, 2016 Irini Fundulaki (FORTH) HOBBIT March 3, 2016 1 / 9
- 2. Voluminous Datasets 1 1http://www.ibmbigdatahub.com/infographic/four-vs-big-data Irini Fundulaki (FORTH) HOBBIT March 3, 2016 2 / 9
- 3. Large number of Linked Data Technologies 2 2https://cdn.datafloq.com/cms/os_big_data_open_source_tools-v2.png Irini Fundulaki (FORTH) HOBBIT March 3, 2016 3 / 9
- 4. Issues to address for Linked Data Technologies √ Which are the critical steps of the Linked Data lifecycle? √ Which are the current bottlenecks? √ Which are the available market solutions? √ Which are the relevant key performance indicators? √ What is the performance of existing tools w.r.t. relevant performance indicators? Irini Fundulaki (FORTH) HOBBIT March 3, 2016 4 / 9
- 5. Holistic Benchmarking of Big Linked Data (HOBBIT) HOBBIT focuses on Big Linked Data and intends to cover the business-critical steps of the Linked Data lifecycle √ Objectives Gather real Performance indicators and Performance thresholds from industry Develop benchmarks based on real data Provide universal benchmarking platform on standardized hardware to provide comparable results Provide periodic benchmarking challenges and reporting Found independent Hobbit association Irini Fundulaki (FORTH) HOBBIT March 3, 2016 5 / 9
- 6. HOBBIT Benchmarks √ Generation and Acquisition: will test the performance of the systems that implement approaches for extracting RDF data from semi-structured data streams such as sensor data (smart metering) and unstructured data streams (Twitter, RSS feeds, etc.) √ Analysis and Processing linking benchmark will test the performance of instance matching methods for Linked Data analytics benchmark will test the performance of supervised and unsupervised machine learning methods Irini Fundulaki (FORTH) HOBBIT March 3, 2016 6 / 9
- 7. HOBBIT Benchmarks √ Storage and Curation storage benchmark will test the performance storage components and small storage subsystems versioning benchmark will examine how versioning systems perform w.r.t. size of the multi-version repository and the efficiency of cross-snapshot query answering √ Visualization and Services: will test the performance of systems regarding question answering and faceted browsing. Irini Fundulaki (FORTH) HOBBIT March 3, 2016 7 / 9
- 8. HOBBIT Overview Irini Fundulaki (FORTH) HOBBIT March 3, 2016 8 / 9
- 9. HOBBIT Logistics √ Horizon 2020, 3 years duration (01/12/2016–30/11/2018) √ Coordinator: Dr. Axel-Cyrille Ngonga-Ngomo, Institute of Applied Informatics, Germany √ Partners: Institute for applied informatics (InfAI), Germany Fraunhofer IAIS, Germany Foundation for Research and Technology (FORTH), Greece National Center for Scientific Research "Demokritos" (NCSR), Greece iMinds, Belgium USU Software AG, Germany ONTOS AG, Switzerland OpenLink, UK AGT Group R&D GMBH, Germany TomTom, Poland Irini Fundulaki (FORTH) HOBBIT March 3, 2016 9 / 9