Abstract The managing of Database (DB) schema evolution is already a very interesting research domain. The main used mechanism for this management is the versioning. It provides a solution for handling changes in the structure of databases and it is widely used in the context of static, temporal and object oriented databases. With the growing interest in geographical databases, the managing... The managing of Database (DB) schema evolution is already a very interesting research domain. The main used mechanism for this management is the versioning. It provides a solution for handling changes in the structure of databases and it is widely used in the context of static, temporal and object oriented databases. With the growing interest in geographical databases, the managing of the schema evolution of these databases becomes a challenge. We propose, in this article, an approach for managing the schema evolution for GDB using temporal versioning. The proposed approach is based on an object oriented formalism. It presents a set of classes for versioning and others for stamping the different versions. It can resolve some problems such as data duplication, the unnecessary copying of information that does not change and the use of “Nil” values. In our approach, data is never physically converted and it is always accessed through conversion interfaces. Read More Read Less

Abstract In this article we describe the experience applied to opinions classification with textometric tools. Our objective is to show, in one hand, the contribution of cooccurrentional analysis to the classification of opinions polarity compared to the purely lexical proceeds, and, in the other hand, the manner of mutualizing the advantages of the co-occurrences study and the classification methods based on... In this article we describe the experience applied to opinions classification with textometric tools. Our objective is to show, in one hand, the contribution of cooccurrentional analysis to the classification of opinions polarity compared to the purely lexical proceeds, and, in the other hand, the manner of mutualizing the advantages of the co-occurrences study and the classification methods based on supervised learning. Relying on a corpus of parliamentary debates issued from the third edition of Text Mining Challenge 2007 we present the results obtained from the characteristics selection according to the textometric analysis. Read More Read Less

Abstract Data reduction is an important pre-processing step to both supervised and unsupervised machine learning problems. In this paper, we investigate, in a first part, the two existing strategies for data reduction which are feature selection (FS) and dimensionality reduction (DR). In a second part, we study the impact of different data reduction methods on supervised machine learning in terms of... Data reduction is an important pre-processing step to both supervised and unsupervised machine learning problems. In this paper, we investigate, in a first part, the two existing strategies for data reduction which are feature selection (FS) and dimensionality reduction (DR). In a second part, we study the impact of different data reduction methods on supervised machine learning in terms of classification accuracy and computational costs. In fact, we compare, in the one hand, the generated subsets of attributes by filter and wrapper algorithms as well as new variables constructed by two variants of a DR method. In the other hand, we compare the classification achieved on initial data set, reduced data sets and also on successively reduced size of the considered data sets. Read More Read Less

Abstract All modern PCs are equipped with an extraordinarily powerful graphic card based on a GPU “Graphic Process Unit”. A component that must be present to be able to animate the 3D games. But, nowadays we use this component to relieve the CPU and thus by asking the GPU to perform heavy parallel computation and quite important applications such as image... All modern PCs are equipped with an extraordinarily powerful graphic card based on a GPU “Graphic Process Unit”. A component that must be present to be able to animate the 3D games. But, nowadays we use this component to relieve the CPU and thus by asking the GPU to perform heavy parallel computation and quite important applications such as image processing and video encoding. This use of GPU has opened a new avenue for improving the performance of multimedia applications. This is the idea behind CUDA: a library of C programming language written by NVIDIA to exploit the capabilities of calculating GeForce through their drivers. While traditional parallel programming libraries will share the work on these CPU on a machine or even a cluster of machines, with planning software, CUDA intends to exploit the extraordinary resources of the GPU graphics processing. The aim of this work is the use of graphics processors to perform general computation usually performed by the central processor CPU. Read More Read Less

Abstract This paper proposes a new method of audio indexing and identification. This new method called CASIA (Calculation of Similarity Audio) is composed by two steps: indexing and identification. In the indexing step, each audio document is pre-processed before the extraction of a set of basic audio descriptors, which characterize the temporal and the spectral information. In the identification step, a... This paper proposes a new method of audio indexing and identification. This new method called CASIA (Calculation of Similarity Audio) is composed by two steps: indexing and identification. In the indexing step, each audio document is pre-processed before the extraction of a set of basic audio descriptors, which characterize the temporal and the spectral information. In the identification step, a query subsequence is pre-processed also and a set of audio descriptors is extracted. The comparison between query descriptors and descriptors of the database allows to generating the interference wave. This wave is used to calculate the similarity rate between the query and documents of database. Read More Read Less