INFORE is a new H2020 project on extreme-scale interactive analytics and forecasting, starting January 2019. At an increasing rate, industrial and scientific institutions need to deal with massive data flows streaming in from a multitude of sources. For instance, maritime surveillance applications combine high-velocity data streams, including vessel position signals emitted from hundreds of thousands of vessels across the world and acoustic signals of autonomous, unmanned vessels; in the financial domain, stock price forecasting and portfolio management rely on stock tick data combined with real-time information sources on various pricing indicators; at the fight against cancer, complex simulations of multi-cellular systems are used, producing extreme-scale data streams in an effort to predict the effects of drug synergies on cancer cells. In these applications, the data volumes are expected to dramatically grow in the future. Processing this data often requires not only using an HPC infrastructure, but also having data scientists, who are typically not expert programmers, program complex workflows, with a vast number of parameters to tune through time-consuming repeated programming and testing. INFORE will address these challenges and pave the way for real-time, interactive extreme-scale analytics and forecasting. The ability to forecast, as early as possible, a good approximation to the outcome of a time-consuming and resource-demanding computational task allows to quickly identify undesired outcomes and save valuable amount of time, effort and computational resources, which would otherwise be spent in vain. Consider, for example, the ability to forecast the outcome of a complex multi-cellular system simulation for tumor evolution, without the need to wait for the simulation to be completed. INFORE will also design and develop a flexible, pluggable, distributed software architecture that is programmable and set up by graphical data processing workflows. The INFORE prototype will be tested on massive real-world data from the life sciences, financial and maritime domains.
Track&Know is a H2020 project running since January 2018. Its mission is to research, develop and exploit a new software framework that aims at increasing the efficiency of Big Data applications in the transport, mobility, motor insurance and health sectors. Stemming from industrial cases, Track&Know will develop user friendly toolboxes that will be readily applicable in the addressed markets, and will be also investigated in additional domains through liaison activities with running ICT-15 Lighthouse projects. Track&Know integrates multidisciplinary research teams from Mobility Data management, Complex Event Recognition, Geospatial Modelling, Complex Network Analysis, Transportation Engineering and Visual Analytics to develop new models and applications. Track&Know recognizes that Big Data penetration is not adequately developed in niche markets outside the traditional verticals (e.g. Finance) and so the Track&Know Toolboxes will be demonstrated in three real-world Pilots using datasets from niche market scenarios to validate efficiency improvements. Performance and impact benchmarks are elaborated and will be documented during pilots deployment. The Track&Know consortium is composed by complementary partners, coming from addressed research, technological and commercial domains, that have a proven track record of high quality research capacity. Thus, the carefully structured workplan, embodies a holistic approach towards meeting the Track&Know objectives and delivering market-relevant outcomes of significant exploitation potential. The CER team will be responsible for the complex event recognition and forecasting technology that will be developed in Track&Know.
datACRON (Big Data Analytics for Time Critical Mobility Forecasting) was a H2020 project running from 2016 to 2019. It targets at introducing novel methods to detect threats and abnormal activity of very large numbers of moving entities in large geographic areas. The datACRON vision is to advance the management and integrated exploitation of voluminous and heterogeneous data-at-rest (archival data) and data-in-motion (streaming data) sources, so as to significantly advance the capacities of systems to promote safety and effectiveness of critical operations for large numbers of moving entities in large geographical areas. The main objectives of datACRON are the development of highly scalable methods for advancing: (a) Real-time detection and forecasting accuracy of moving entities’ trajectories; (b) real-time recognition and prediction of important events concerning these entities, together with (c) a general visual analytics infrastructure supporting all steps of the analysis through appropriate interactive visualizations; (d) advanced processing of data close to the data sources -following the in-situ data processing paradigm, producing streaming data synopses at a high-rate of compression; (e) advanced spatio-temporal data integration and management solutions. The CER team will be responsible for the complex event recognition and forecasting technology that will be developed in dataCRON.
SPEEDD (Scalable ProactivE Event-Driven Decision-making) was a European project that ended on 1-2-2017. SPEEDD developed a prototype for proactive event-driven decision-making: decisions triggered by forecasting events-whether they correspond to problems or opportunities-instead of reacting to them once they happen. Decision-making in SPEEDD was real-time, in the sense that it takes place under tight time constraints, and require on-the-fly processing of Big Data, that is, extremely large amounts of noisy data flooding in from different geographical locations, as well as historical data. The SPEEDD methodology for proactive event-based decision-making comprises the following steps. First, Big Data is continuously acquired from various types of sensor and fused in order to recognise, in real-time, events of special significance. To allow for sub-second recognition, SPEEDD minimizes communication volume by moving as little data as possible from one place to another. Second, the events recognised are correlated with historical information to forecast problems and opportunities that may take place in the near future. Third, the forecast events along with the recognised events are leveraged for real-time operational decision-making. Fourth, visual analytics tools prioritise and explain possible proactive actions, enabling human operators to reach and execute the correct decision. The SPEEDD technology will be tested in proactive traffic management and proactive credit card fraud management. The CER team coordinated the project and contributed to the development of novel techniques for real-time event recognition and forecasting under uncertainty.
REVEAL (REVEALing hidden concepts in social media) was EU FP project, that ended on 1/11/2016. The world of media and communication is currently experiencing enormous disruptions: from one-way communication and word of mouth exchanges, we have moved to bi- or multi directional communication patterns. No longer can selected few (e.g. media organizations and controllers of communication channels) act as gatekeepers, deciding what is communicated to whom and what not. Individuals now have the opportunity to access information directly from primary sources, through a channel we label e'-word of mouth', or what we commonly call 'Social Media'. A key problem: it takes a lot of effort to distinguish useful information from the 'noise' (e.g. useless or misleading information). Finding relevant information is often tedious. REVEAL aimed at discovering higher level concepts hidden within information. In Social Media we do not only have bare content; we also have interconnected sources. We have to deal with interactions between them, and we have many indicators about the context within which content is used, and interactions taking place. A core challenge was to decipher interactions of individuals in permanently changing constellations, and do so in real time. Further to discovering what is being said, REVEAL aimed at determining how trustworthy that information is, based on predicting contributor impact and how much or to what extent all this affects reputation or influence. This allowed to automatically judge the quality and accuracy of content, and predicting future trends with greater accuracy. The core of our work was to reveal hidden social media modalities for the benefit of a better understanding and utilization of the Social Media world. The CER team was responsible for social modality recognition using heterogeneous data streams coming from various types of social media.
AMINESS (Analysis of Marine Information for Environmental Safe Shipping) . An GSRT/ Ministry of Development-funded project ending June 2015. The CER team was responsible for the probabilistic event recognition technology used for the detection of potential hazards related to marine traffic in the Aegean sea, towards reducing the possibility of ship accidents.
USEFIL (Unobstrusive Smart Environments For Independent Living). EU FP7 project ending on November 2014. The CER team was responsible for the probabilistic event recognition technology that was used for the unobtrusive monitoring of elderly people in their smart homes.
PRONTO (Event Recognition for Intelligent Resource Managament). EU FP7 project ended on March 2012. The CER team was responsible for the event recognition technology that was used for city transport management and emergency rescue operation management.