Abstract The SQL Server 2012 has a new feature to improve performance of data warehouse queries called Columnstore index, stores data by columns instead of by rows. This paper discusses the new type of index in some details, what is a column store, how and why improves the speed up and performance of the same queries that processing in SQL Server... The SQL Server 2012 has a new feature to improve performance of data warehouse queries called Columnstore index, stores data by columns instead of by rows. This paper discusses the new type of index in some details, what is a column store, how and why improves the speed up and performance of the same queries that processing in SQL Server when it used this index, and talk a limitation and advantages of this new indexes, then the summarization and resulting is illustrated by example queries. And it will clarify many things on this subject. Read More Read Less

Abstract Next Generation Wireless Networks (NGWN) is a ubiquitous environment that integrates various wireless technologies such as WLAN, 3G, and WiMAX based on IP. One of the most challenging issues of NGWN is providing seamless service continuity while user roaming among different wireless access networks. Context awareness has showed a significant effect to guarantee services continuity. In this paper, we will... Next Generation Wireless Networks (NGWN) is a ubiquitous environment that integrates various wireless technologies such as WLAN, 3G, and WiMAX based on IP. One of the most challenging issues of NGWN is providing seamless service continuity while user roaming among different wireless access networks. Context awareness has showed a significant effect to guarantee services continuity. In this paper, we will evaluate the proposed Context-Aware Management Scheme (CAMS) in the case of video application in terms of different performance parameters such as jitter, packet loss and end-to end delay. Read More Read Less

Abstract The development of context-aware ubiquitous ap-plications has been an emergent subject of several researches in pervasive computing. In different context modelling proposed approaches, we notice that there is still no consensus and their context models are either incomplete or domain specific. This paper aims to summarize these approaches and proposes a context model which should be common to be used... The development of context-aware ubiquitous ap-plications has been an emergent subject of several researches in pervasive computing. In different context modelling proposed approaches, we notice that there is still no consensus and their context models are either incomplete or domain specific. This paper aims to summarize these approaches and proposes a context model which should be common to be used by different context-aware applications, specific to cover the main contextual entities existing in the literature and flexible to allow taking into account new entities specific to a given application domain. To accomplish these requirements, we adopt an ODM (Ontology Definition Meta model) approach to create a high level modelling of context information using ontologies. The main objective of the ODM is to bridge the gap between traditional software tools for modelling and artificial intelligence techniques for making ontologies. Read More Read Less

Abstract The complexity, abundance and diversity of data (text, images, sound and video) are at the origin of new software solutions for their exploitation. These solutions make use of learning techniques. Among of these techniques, we distinguish learning through co-training, which turned out to be a very exploited by researchers thanks to its low sensitivity to noise, robustness and rapidity. However,... The complexity, abundance and diversity of data (text, images, sound and video) are at the origin of new software solutions for their exploitation. These solutions make use of learning techniques. Among of these techniques, we distinguish learning through co-training, which turned out to be a very exploited by researchers thanks to its low sensitivity to noise, robustness and rapidity. However, the classical variant of co-training showed deficiencies in the classification of complex data particularly in terms of adaptability. In this paper, we present a new approach of semi-supervised learning by Co-training technique to generate a generic classifier from a representative database (MGLP) which is automatically labeled online. Robustness and rapidity, of our approach, to classify complex data were demonstrated through a comparison between the classical variant and the new one in terms of rate and time. Read More Read Less