A distributed data-mining
software platform for
EXTReme dAta Across the
Compute conTinuum

Delivering a data-driven open-source platform integrating cloud, edge and HPC technologies for trustworthy, accurate, fair and green data mining workflows for high-quality actionable knowledge

Objectives

Enable the development of complex and secure data mining workflows

Develop novel data-driven orchestration mechanisms to efficiently deploy and execute data mining workflows

Deliver the EXTRACT software platform and demonstrate its benefits in two use cases

Fully exploit the performance capabilities of the compute
continuum to effectively address extreme data characteristics (high
volume, variety, velocity, veracity) holistically

Foster the adoption of EXTRACT technology by industrial and academic communitie

Use cases

Personalized Evacuation Routing (PER) System

A Personalized Evacuation Routing (PER) System will serve to guide citizens in an urban environment (the city of Venice) through a safe route in real time. The EXTRACT platform will be used to develop, deploy and execute a data-mining workflow to generate personalized evacuation routes for each citizen, displayed in a mobile phone app, by processing and analysing extreme data composed of Copernicus and Galileo satellite data, IoT sensors installed across the city, 5G mobile signal, and a semantic data lake fusing all this information.

Transient Astrophysics with a Square Kilometer Array Pathfinder  (TASKA)

The Transient Astrophysics with a Square Kilometer Array Pathfinder (TASKA) case will use EXTRACT technology to develop data mining workflows that effectively reduce the huge amount of raw data produced by NenuFAR radio-telescopes by a factor of 100. This will allow the populating of high-quality datasets that will be openly accessible to the astronomy community (through the EOSC portal) to be leveraged for multiple research activities.

Simulating Movement to Test Learning Strategies in the PER use Case

Simulating Movement to Test Learning Strategies in the PER use Case

Implementing the PER  system and creating a Multi-Agent Reinforcement Learning (MARL) model requires a preliminary phase that simulates the possible movements of people in an emergency situation to properly train the model.. A true-to-life simulator offers a safe and controlled platform for testing learning strategies in different situations.

URV presents at MIDDLEWARE 2023

URV presents at MIDDLEWARE 2023

BSC partner Universitat Rovira i Virgili (URV) presented its paper, "Glider: Serverless Ephemeral Stateful Near-Data Computation" on 13 December 2023. Researcher Daniel Pons delivered the presentation as part of Session1A- Serverless. The publication is available here...

9th Int´l WS on Serverless Computing

9th Int´l WS on Serverless Computing

Partner URV participated in the 9th International Workshop on Serverless Computing (WoSC9). This workshop brought together researchers and practitioners to discuss their experienes and thoughts on future directions for the field. More information on the workshop is...