1. Introduction
Although important advances in multimedia mining have been achieved, many old problems remain and are ever major research areas. These research areas include multimedia semantics and user interaction modeling.
The first major research area, multimedia semantics [Citation1], makes sense out of multimedia by semantically linking low level features (e.g. color, texture, shape) and high level features (e.g. people, adoration, actions). A key approach is the extraction of semantics from multimedia with limited manual annotation efforts, based, on a semi-supervised learning process. Another key approach is the automatic extraction of text from speech accompanying video, or recognition of text written on images.
The second major research area, user interaction modeling, consists of tracking and analyzing user actions on multimedia information, in order to deduce their interestingness. It includes the personalization of queries and the identification of the frequent pattern of actions [Citation2].
2. Special issue articles
Some papers of the special issue are based on the original submissions to the 7th International Workshop on Multimedia Data Mining (MDM-2006, held in Philadelphia, August 20–23, 2006, USA). Other papers have been invited, on the basis of the smart expertise of their authors. The purpose of the workshop was to bring together researchers, developers and practitioners from academia and industry to discuss challenges in multimedia mining. In this special issue, we sought contributions from a wide range of theoretical and application areas. When selecting the contributions to be presented in this issue, we aimed at providing a good balance of research areas and contributions to multimedia mining. The selected articles fall into broad categories that reflect the variety of research directions in multimedia mining. The purpose of this special issue is to present some contributions in the domain covering a broad range of topics. The special issue is composed of five papers that investigate video, images and processes. It is composed of:
-
Multimedia enriched ontologies for video digital libraries, by Marco Bertini, Alberto Del Bimbo and Carlo Torniai.
-
Learning multiple linear manifolds with self-organizing networks, by Huicheng Zheng, Pádraig Cunningham and Alexey Tsymbal.
-
An efficient spatial semi-supervised learning algorithm, by Ranga Raju Vatsavai, Shashi Shekhar and Thomas E. Burk.
-
Relation rule mining, by Mehdi Adda, Rokia Missaoui and Petko Valtchev.
-
A time Petri net-based approach for synchronization, analysis and management of multimedia scenarios, by A. Ghomari and M. K. Rahmouni.
References
- Djeraba , C. , Mongy , S. and Bouali , F. 2006 . “ Video usage mining ” . In Chapter of Encyclopedia of Multimedia , Edited by: Furht , B. Springer .
- Zhao , R. and Grosky , W. 2002 . Negotiating the semantic gap: from feature maps to semantic landscapes . Pattern Recognition , 35 ( 3 ) : 51 – 58 . March