The very large amounts of digital data generated dynamically by end users, communities and devices require ways for efficient representation and organization, in order to be used in applications and services. The well-studied approaches of analyzing data in terms of objects, concepts, tags and other isolated entities are often insufficient because of the fact that they do not take into account important properties of data such as spatio-temporal aspects. Event representation has emerged as a more natural abstraction of happenings in the real world, matching the way humans think in terms of events and entities. Methods for event recognition, therefore, have been attracting considerable attention. Consider, for example, recognition of trends, given user contributions in social web 2.0 applications; recognition of attacks on nodes of a computer network, given the exchanged TCP/IP messages; recognition of suspicious trader behaviour, given the transactions in a financial market; and recognition of various types of cardiac arrhythmia, given electrocardiograms.
Events can be rather complex entities and, in addititon to that, most current recognition approaches exploit high-level domain-specific contextual information; therefore, efficient structures for representing and processing such kind of information are needed, usually in the form of ontologies and rule-based systems. Temporal aspects are inherent in the definition of events, whereas real-time techniques—applied to very large data streams—are necessary for providing time-critical information. Events are important for multimedia analysis and annotation—in dynamic sound and video signals human behavior and activities are better described and exploited in terms of events. These are but a few of the requirements and aspects of event recognition addressed by the papers published in this Special Issue.
THE ARTICLES
The call for papers of the Special Issue received a strong response from various communities, including Artificial Intelligence, Distributed Systems, Multi-Agent Systems, and Multimedia Analysis. A total of 15 manuscripts were submitted for consideration. Of these submissions, six papers were accepted, following a rigorous two-stage review process coordinated by the guest editors. The accepted papers may be classified in the following categories.
Knowledge-Based Approaches
The first part of the Special Issue consists of two articles that use ontologies of contextual information in order to enhance event recognition accuracy. Anicic et al., in the article entitled “Real-Time Complex Event Recognition and Reasoning – A Logic Programming Approach,” present a knowledge-rich complex event processing engine which, apart from events, also processes contextual knowledge. The engine includes a rule-based language for pattern matching over event streams. Devaurs et al., in their article “Exploiting the User Interaction Context for Automatic Task Detection,” present an approach for detecting the tasks a user is performing on his desktop by using an ontology-based user interaction context model. The authors empirically show that their ontology-based approach enhances task detection performance.
Events in Multimedia
The second part of the Special Issue includes articles from the field of Multimedia Analysis. Two papers concern approaches on complex event recognition from video content—“activity recognition” in the terminology of these papers. Hu and Boulgouris, in their article “Visual Recognition of Events and Activities based on Momentum of Motion Energy Mass,” propose a recognition system that combines the apparent moving areas and the velocity of the associated movement. Consideration is also given to the movement that is not taking place in the projection plane of the camera. Voulodimos et al., in the paper entitled “Improving Multi-Camera Activity Recognition by Employing Neural Network Based Readjustment,” propose a method for enhancing recognition in environments where problems like occlusions, outliers, and illumination changes occur. In order to address these problems, multiple cameras are used, while a neural network-based readjustment mechanism dynamically corrects erroneous classification results for image sequences.
The third article of this part concerns audio event recognition. Ramona et al., in “A Public Audio Identification Evaluation Framework for Broadcast Monitoring,” present a framework for the evaluation of audio identification techniques—a type of audio event recognition. The proposed framework concerns use-cases in which audio excerpts have to be detected in a radio broadcast stream that include a large variety of audio distortion.
Applications of Event Recognition
The third part of the Special Issue includes a paper reporting an event recognition application. Ranathunga et al., in “Identifying Events Taking Place in Second Life Virtual Environments,” present a framework for recognising events in the Second Life virtual world. Event recognition, in this case, consists of extracting low-level spatio-temporal data and identifying the embedded high-level domain-specific information.
The Special Issue demonstrates the broad diversity of the approaches on event recognition. This is a topic of active research in various fields, such as Artificial Intelligence, Multimedia Analysis, Distributed Systems, Database Systems, and Software Engineering. It is clear that there is a lot to be gained by bringing these research communities closer together and combining their approaches. We hope that this Special Issue is another step towards this direction.
BIOGRAPHIES OF THE EDITORS
Dr. Alexander Artikis is a Research Associate in the Institute of Informatics & Telecommunications at NCSR “Demokritos” in Athens, Greece. He holds a PhD from Imperial College London on the topic of norm-governed multiagent systems. His research interests lie in the areas of distributed Artificial Intelligence, temporal representation and reasoning, Artificial Intelligence and law, and description logics. He has published papers in related journals and conferences, such as the Artificial Intelligence Journal, the ACM Transactions on Computational Logic, and the Journal of Logic & Computation. He is currently working on the EU FP7 PRONTO project, being responsible for the event recognition work-package. In the past he has worked for several international and national projects, including the highly successful EU FET ALFEBIITE project. Dr. Artikis has co-organized and served as a member of program committees for several conferences and workshops.
Dr. Ioannis (Yiannis) Kompatsiaris is a Senior Researcher (Researcher B) with the Informatics and Telematics Institute. His research interests include semantic multimedia analysis, indexing and retrieval, Web 2.0 content analysis, knowledge structures, reasoning and personalization for multimedia applications. He received his Ph.D. degree in 3-D model based image sequence coding from the Aristotle University of Thessaloniki in 2001. He is the co-author of 49 papers in refereed journals, 27 book chapters, 4 patents and more than 150 papers in international conferences. He is co-editor of the book Semantic Multimedia and Ontologies: Theory and Applications, the guest editor of four special issues, and he has served as a program committee member and regular reviewer for a number of international journals and conferences. He has been the co-organizer of various conferences and workshops, such as the ACM CIVR, WIAMIS and SSMS. He is the coordinator of the WeKnowIt – Emerging, Collective Intelligence for personal, Organisational and Social use European Integrated Project. Recently, he has been appointed as Chair of the Technical Committee 14 of the International Association for Pattern Recognition (IAPR-TC14, “Signal Analysis for Machine Intelligence”). He is a Senior Member of IEEE and a member of ACM.
Thomas Winkler is a researcher at the Fraunhofer Institute for Intelligent Analysis and Information Systems (IAIS) in Germany. He received his degree in Electrical Engineering in 2005 from the RWTH Aachen University, Germany. The focus of his studies were multimedia signal processing and classification—in particular, approaches for blind source separation. In 2006 he joined the Fraunhofer IAIS in Sankt Augustin, working in several national and European research projects (SHARE, MOVEON, PRONTO, etc.) in the area of speech analysis, dialogue interaction, and event recognition for decision support. In 2010 he co-organised the EVENTS workshop at the SETN conference in Athens. His current research interests are in various aspects of multimedia pattern recognition from audio-visual similarity to robustness in automatic speech recognition.
Dr. Phivos Mylonas was born in Athens in 1978. He obtained his Diploma in Electrical and Computer Engineering from the National Technical University of Athens (NTUA) in 2001, his Master of Science (M.Sc.) in Advanced Information Systems from the National & Kapodestrian University of Athens (UoA) in 2003 and his PhD at the former University (NTUA) in 2008. He is currently a Researcher at the Image, Video and Multimedia Laboratory, School of Electrical and Computer Engineering, Department of Computer Science of the National Technical University of Athens, Greece. His research interests include multimedia information retrieval, knowledge-assisted multimedia analysis, issues related to multimedia personalization, user adaptation, user modeling and profiling, visual context representation and analysis. He has published articles in 26 international journals and book chapters, he is the author of 45 articles in international conferences and workshops, he has edited eight books and is a guest editor of five international journals, he is a reviewer for 10 international journals and has been actively involved in the organization of 23 international conferences and workshops. He has been a member of the Technical Chamber of Greece since 2001, a member of the Hellenic Association of Mechanical & Electrical Engineers since 2002 and a member of W3C since 2009, and he has been an IEEE member from 1999 to 2010 and an ACM member from 2001 to 2010.
Acknowledgments
This Special Issue has been edited by members of consortia of two related EU FP7 projects: PRONTO (http://www.ict-pronto.org/) and WeKnowIt (http://www.weknowit.eu/). In brief, PRONTO proposes a methodology for fusing data from various sources, analysing it to extract useful information in the form of events and making them available for intelligent resource management. WeKnowIt exploits multiple layers of intelligence from user-generated content in order to transform the large-scale and poorly structured Social Media to meaningful topics, entities, points of interest, social connections and events.