#	Pagina
attuale pagina	/open-h2020/projects/216142/results.html

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 1 - MAGNETO (Multimedia Analysis and Correlation Engine for Organised Crime Prevention and Investigation)

Teaser

Summary

Investigators all over Europe are very often confronted with a variety of complex cases in which huge amounts of heterogeneous data have to be analysed for the prevention, investigation and prosecution of criminal offences. The analysis of such data volumes frequently exceeds the capacity of the police teams. Moreover, in some cases, both the workload and time constraints can be extreme, such as in a sequence of terrorist attacks which, from the LEAs\' part, involve pre-incident intelligence analysis, real-time incident processing and post-incident evolution prediction linked potentially to other threats or subsequent attacks. In this light, MAGNETO aims to enhance the capabilities of LEAs in managing, investigating, correlating and reasoning upon huge volumes of heterogeneous and disjoint multimedia data, in order to accomplish highly-efficient crime prevention and investigation. Specifically, MAGNETO provides powerful, intelligent, intuitive and modular analytics tools through which investigators can receive comprehensive answers to their queries and discover relations between data more easily, gaining better insights. The cornerstone of MAGNETO is, accordingly, the definition and specification of the common representational model for data, together with the development of advanced semantic reasoning and augmented intelligence tools to support LEA officers, integrated in a common platform that allows storage, handling and processing of the collected multi-source information.

MAGNETO\'s objectives are to:
- Augment the capabilities of Law Enforcement Agencies (LEAs) in managing, investigating, correlating and reasoning from huge volumes of heterogeneous and disjoint multimedia data.
- Develop dedicated technical breakthroughs to anticipate both the growth in volume and types of data, and the resources and evolution of crime and criminality.
- Design and compose an open framework and develop powerful tools to deliver increased innovation capacity in Europe, and more efficient information exchange between operators and other users.

Work performed

During the reporting period, MAGNETO identified appropriate scenarios, use cases and end-user requirements. The initial MAGNETO architecture has been designed while a security-related ontology has been designed and analyzed. MAGNETO has also focused on extracting information from video and from textual data in English language. Furthermore, pre-processing of sensor data from heterogeneous capturing devices has also been performed, resulting in the extraction of deep-learning features for the purposes of person identification and recognition. In addition, MAGNETO supports Automatic Language Translation with the development of Machine Translation (MT) engines. A reasoning tool based on logical and probabilistic reasoning is being developed to create new knowledge based on rules that need to be defined in cooperation with LEA users, while internal tools to fuse trajectories, as well as persons, are designed.
Two different visualization tools have been developed; an immersive prototype able to visualize geo-referenced information in a Virtual Reality (VR) environment, and a web Human-Machine Interface (HMI) able to interact with the different MAGNETO services and stored data. In order to deliver training modules of the MAGNETO platform, a comprehensive questionnaire framework was established, while three formal methodologies for the development of the training solutions have been identified: MAGNETO platform manual, webinars, Face to Face (F2F) workshops. It is worth mentioning that MAGNETO is a project that is aligned with ethical and legal guidelines.
Interconnected to the research and technical activities, MAGNETO also conducted a large number of dissemination and communication actions regarding the project concept and goals. Project presentations to various stakeholders were performed, while MAGNETO partners attended various related events and conferences. MAGNETO has also become a member of ongoing sister projects including COPKIT, CONNEXIONS, TENSOR, ANITA SPIRIT, ASGARD, TITANIUM, iLEAnet and I-LEAD.

Final results

The ambition of the MAGNETO project is to empower LEA analysts and investigators with tools and infrastructures not commercially available today, to handle the petabytes of data that must be processed during an investigation. The solutions, tools and infrastructures offered by MAGNETO will provide important advances beyond the state-of-the art, which include:
-Multimedia analysis: MAGNETO aims to perform content and context metadata extraction, data indexing, data fusion and efficient search and retrieval.
-Semantic information fusion: MAGNETO aims to build a comprehensive Semantic Information Fusion system based on recent advances, to deploy it at the test sites, and to implement an ontology modelling the classes and their interrelations for the data streams arriving at the information fusion stages.
- Text mining, data mining, pattern recognition and machine learning: In MAGNETO, semantic notations and real-time machine learning algorithms will enable efficient processing and analysis in terms of speed, variety and different types of content.
- Heterogeneous data exchange and sharing: MAGNETO\'s ontology will allow knowledge exchange between different LEA systems.
- Augmented intelligence: MAGNETO investigates novel immersive Human-Machine Interfaces (HMIs), based on web and virtual reality technologies, creating a multidimensional environment that allows better situational awareness.

The MAGNETO project is expected to deliver the following results:
- Evidence collection platform employing heterogeneous data mining and multimedia content indexing making it available for further processing.
- Sophisticated representational model for representing knowledge in an open, standardized manner to unlock the discovery of relations and enable interoperability.
- Advanced correlation engine, which will facilitate automatic discovery of new, unsuspected relations within big datasets.
- Threat prediction engine by semantic reasoning, which will enable trends analysis and threat level assessment.
- Augmented intelligence tools that will deliver immersive and interactive HMIs to enhance situation awareness.
- Results transferability framework.
- Legal and ethical assessment framework for examining the impact of MAGNETO and ensuring its compliance to ethical standards and legislation.
- Training curricula for the use of MAGNETO tools.

MAGNETO brings a set of substantial impacts including:
- Improvement of investigation capabilities of the LEAs, by providing tools that reduce the time to solve crimes.
- Enhanced prediction and prevention of terrorist endeavors.
- Better identification and understanding of criminal activities.
- Better capabilities to prevent, investigate, and mitigate criminal acts.
- Compliance with the Privacy and Ethical directives of EU.