Search

A set of actors may take part in numerous categories of real-life social networks. For instance an actor may take part in a mobile social network and a friendship network. The interactions between actors in the mobile social network may not be reproduced in the friendship one. We accept that overlaying networks with different kinds of interactions assists on detecting interesting patterns. The main goal of this paper is to overlay two or many social networks with different kinds of social activities in order to mine consistent communities that could not be appeared in a single social network. In this respect, we introduce a new community detection approach based on injecting the perspective communities from a mobile social network into an initial friendship network within a sequence of time windows. In other words, during each time window, we constructed an augmented friendship social network using the semantic information extracted from perspective communities and from it to detect communities that may belong to several networks in a given time period. We demonstrate the abilities and potential of our approach by evaluating it using real datasets and comparing it with two previous methods.

The Moroccan health sector is now inspired by other industries to take advantage of the benefit of cloud computing in terms of cost and flexibility, without compromising data security. Indeed, the medical field has an exponential number of data that requires a developed infrastructure and a very high storage and archiving capacity, so the process of processing the patient's information is slow and can return erroneous results. Considering the increasing demands placed on healthcare providers for infrastructure, systems and support, cloud computing technology can greatly improve the level of healthcare services; it will be the appropriate solution to meet the needs for care coordination, collaboration and communication between the different actors (doctors, patients, administrators, developers…) in the health field. In order to implement the project “family health book”, we need an architecture model that will be based on cloud computing while offering QoS (Quality of Service) required, to meet this need, we suggest in this article a solution based on the use of a cloud broker open source in an inter-cloud environment, in order to allow the technological renewal of our infrastructure while reducing both the time of access to the services requested by the users, the dissipation of the energy and the number of datacenters installed. Evaluation results demonstrated that our architecture has achieved good performances.

This paper presents a case of study of a virtual learning environment based on Agile Learning Design approach and integrating learner-centered design. It aims to give the foundation of a framework that gives extensive attention at each stage of the design process to the end user — which is in our case the learners-to improve teaching effectiveness, facilitate learning among learners, encourage long life learning and maximize motivation as well as reducing the dropout rates.

Social networks are dynamic topologies with a set of nodes playing a significant role in the information flow transmission within the network. The appearance of new links in the network may change the role of nodes and conduct to a rapid information flow. For example an individual may belong to an online social network and a friendship network. The relationships between individuals in the mobile social network may not be reflected in the friendship one. We have confidence that overlapping networks with several kinds of interactions provide a fast propagation of information and helps to identify critical and/or non-critical nodes. This article aims to extend our previous work by injecting the perspective communities from a mobile social network from a sequence of time windows in a friendship social network. Centrality and distance measures were applied to validate the impact of perspectives communities on the quality of information flow.

Extracting knowledge from text data and taking its full advantage has been an important way to reduce its computation and accelerate processing, especially for large amounts of data. Thus, different approaches and methodologies for modeling and representing textual data have been proposed. In this paper, a graph-based approach for automatic indexing of unstructured data from an Arabic corpus has been proposed. First, each document in the collection is represented by a graph. After the generation of document graph, term weighting is computed to estimate the relevance of a term to the document. The graph representation offers the advantage that it allows for a much more expressive document modeling than the standard bag of words approach, and consequently, it improves classification performance. Experimental results show that the graph based indexing method is a promising approach for semantic and contextual indexation, and outperforms statistical based method (TFIDF) by 12% in F-measure.

Day by day, the amount of digital data grows exponentially. Furthermore, these data come from heterogeneous content and structure resources. Both factors are the major handicap that limits the correct functioning of information systems. Being on the same wavelength, researchers have fixed as a challenge to reach targeted information retrieval which allows an easy access to the desired resource then return relevant and correct outcomes corresponding to a specific request. Therefore, semantic annotation is chosen as a paramount solution that may handle all these requirements, while ensuring a good understanding of document content then exploit and share it readily afterward. In this paper, we treat semantic annotation of documents using multi-ontologies, by stressing the effect of that on the relevance and the quality of semantic annotation results. We present a survey of different multi-ontologies based semantic annotation approaches and conclude with a synthesis of the study made.

In this paper we present a system for offline recognition cursive Arabic handwritten text based on Hidden Markov Models (HMMs). The system is analytical without explicit segmentation used embedded training to perform and enhance the character models. Extraction features preceded by baseline estimation are statistical and geometric to integrate both the peculiarities of the text and the pixel distribution characteristics in the word image. These features are modelled using hidden Markov models and trained by embedded training. The experiments on images of the benchmark IFN/ENIT database show that the proposed system improves recognition.

Documents indexing is the main step in a conventional document classification or information retrieval framework. This study aims to highlight the influence of features' type on the efficiency of a classification system. Empirical results on Arabic dataset reveal that the choice of extracted feature's type has a significant impact on conserving semantic information and improving classification accuracy, especially with the morphological complexity of the Arabic language. Precision, recall, F-measure and accuracy are the metrics adopted to compare the efficiency of the proposed indexing system.

In this paper we present a system for offline recognition cursive Arabic handwritten text based on Hidden Markov Models (HMMs). The system is analytical without explicit segmentation used embedded training to perform and enhance the character models. Extraction features preceded by baseline estimation are statistical and geometric to integrate both the peculiarities of the text and the pixel distribution characteristics in the word image. These features are modelled using hidden Markov models and trained by embedded training. The experiments on images of the benchmark IFN/ENIT database show that the proposed system improves recognition.

This paper is devoted to describe a preliminary draft of our approach that aims to identify and track learners' learning styles based on their behavior and actions during a MOOC then to provide them with personalized recommendations based on their learning styles. Massive Open Online Courses are attracting a debate in the research community about their influence in online education. Indeed, with their advent, we are assisting to a substantial expansion of online learning with new advantages such as: massiveness, openness, democratization of learning, etc. However, it raises particular issues related to the dropout rates and the heterogeneity of massive learner's cohorts. In this approach, we use neural networks for the identification and tracking of learner's learning styles in MOOCs so as to increase learners' engagement and satisfaction. The purpose of this paper is to examine the point of view of literature, the dropout issue and solution to integrate an adaptive recommendation system with MOOC.

This paper presents a machine vision system for automatic recognition of plant leaves diseases from images. The proposed system is based on serial combination technique of two SVM classifiers. The first classifier uses the color to classify the images; it considers, at this phase, that the diseases with similar or nearest color belonging to the same class. Then, the second classifier is used to differentiate between the classes with similar color according to the shape and texture features. The tests of this study are carried out on six classes of diseases including three types of pest insects damages (Leaf miners, Thrips and Tuta absoluta) and three forms of pathogens symptoms (Early blight, Late blight and Powdery mildew). The results of the study show the advantages of the proposed method compared to the other existing methods.

This study presents a multiple classifier system for automatic recognition of the damages and symptoms on plant leaves from images. The proposed approach is based on parallel combination of two kinds of classifiers, one is a neural network classifier that uses texture, color and shape features to distinguish between the damages and symptoms, then the other is a support vector machine (SVM) classifier that uses texture and shape features. In order to design our system, we have based on some existing approaches in the field that adopt a single classifier. The tests of this study were carried out on six classes including the damages of three pest insects (Leaf miners, Thrips and Tuta absoluta) and symptoms of three fungal diseases (Early blight, Late blight and Powdery mildew). The experimental results show the efficiency of our approach compared to the pervious approaches based on single classifier. The proposed approach is more effective and has the highest rate of recognition.

The proliferation and widespread of social web provide rich data about peoples' activities. For instance, geographical position and timestamps of each event can be obtained thanks to mobile devices and location-based services. Basically it sounds well that people, who used to be frequently at the same place and the same time period, are very likely to be socially related. Therefore, it is worth-noting that tracking user's position over time may help discovering interesting patterns, which can be used to enhance community structure. The goal of this paper is to track and collect users' spatiotemporal activities events in order to unveil homogeneous groups that could not appear in a unique social network. In this respect, we propose a community detection approach from spatiotemporal aspects. First, we use ST-DBSCAN clustering algorithm to identify clusters corresponding to temporary communities that we called perspective. Then, we apply different random graph models to establish relationships between pairs of individuals. This real-time detecting algorithm is done to track the dynamic evolution of communities within a sequence of time windows. Finally, we test our proposed method on a real-life datasets in order to analyze the soundness and feasibility of this technique.

The proposed approach aims at designing an automatic recognition system of the damages and symptoms on plant leaves. It is based on serial combination of two neural networks classifiers. The first classifier uses the color to differentiate between classes. Indeed, at this phase, the damages and/or the symptoms that have a similar or a nearest color are considered to belong to the same class. Then, the second classifier is used to differentiate between classes with similar color according to the shape and texture features. The approach is tested on four classes, including the damages of two kinds of pest insects (Leaf miners and the caterpillar Tuta absoluta), and the symptoms of two fungal diseases (Downy mildew and internal Powdery mildew). The experimental results indicate that the proposed approach would be interesting to use as means of diagnosis and phytosanitary problem recognition from images of damages and symptoms.

Page segmentation into text and non-text elements is an essential preprocessing step before optical character recognition (OCR) operation. In case of poor segmentation, an OCR classification engine produces garbage characters due to the presence of non-text elements. This paper presents a method to separate the textual and non textual components in document images using a graph-based modeling and structural analysis. This is a fast and efficient method to separate adequately the graphical and the textual parts of a document. We have evaluated our method on two well-known subsets: the UW-III dataset and the ICDAR 2009 page segmentation competition dataset. Comparisons are led with two methods of state-of-the-art, these results showing that our method proved better performances in this task.

Libraries contain huge amounts of Arabic printed historical documents which cannot be available on-line because they do not have a searchable index. The word spotting idea has previously been suggested as a solution to create indexes for such a collection of documents by matching word images. In this paper we present a word spotting method for Arabic printed historical document. We start with word segmentation using run length smoothing algorithm. The description of the features selected to represent the words images is given afterwards. Elastic Dynamic Time Warping is used for matching the features of the two words. This method was tested on the Arabic historical printed document database of Moroccan National Library (MNL).

Page segmentation into text and non-text elements is an essential preprocessing step before optical character recognition (OCR) operation. In case of poor segmentation, an OCR classification engine produces garbage characters due to the presence of non-text elements. This paper presents a method to separate the textual and non textual components in document images using a graph-based modeling and structural analysis. This is a fast and efficient method to separate adequately the graphical and the textual parts of a document. We have evaluated our method on two well-known subsets: the UW-III dataset and the ICDAR 2009 page segmentation competition dataset. Comparisons are led with two methods of state-of-the-art; these results showing that our method proved better performances in this task.

The goal of this paper is to examine the importance of knowledge management and semantic web in tourism. We will primarily aim to present the concepts of knowledge management, knowledge management systems, semantic web and ontologies. We also provide the importance of knowledge management in tourism, which gave the real benefits to be derived from the use semantic-web based ontologies for travelers. Finally, the last part is devoted to present tourism ontology for Morocco and architecture of the KMS we conceived.

Digital watermarking is a technology for embedding digital information in digital content (audio, images, video…). Recently, the watermarking is introduced as a tool to improve the security of identity documents. In this paper, we propose a watermarking algorithm operating in discrete wavelet transform (DWT) using mixed scales representation and quantization index modulation (QIM) of dominant blocks. Dominant blocks of the DWT with mixed scales represent the textured and contours zones of a portrait image. They are used for improving watermark invisibility and robustness. The DWT of Faber-Schauder (FSDWT) is firstly applied to image with mixed scales then the dominant blocks are selected. The watermark is embedded into the dominant blocks to obtain the watermarked image. Experimental results demonstrate that the proposed method is robust to attacks such as compression, blur, and noise.

Adaptive hypermedia systems traditionally focus on providing personalized learning services. This feature involves the search for content tailored to a given learner. To accomplish this goal, sophisticated search mechanisms are required, particularly the semantic tools. The goal of this paper is to examine the issue of the semantic web and Learning Objects in Adaptive Hypermedia Systems, to assemble and generate, from a repository of Learning Objects, a course adapted to the needs, goals and preferences of a given learner.

The classification of remotely sensed images knows a large progress seen the availability of images of different resolutions as well as the abundance of the techniques of classification. Moreover a number of works showed promising results by the fusion of spatial and spectral information. For this purpose we propose a methodology allowing to combine this two information to refine an SVM classification, The approach uses Haralick texture features extract from GLCM as space descriptors to be combined with spectral information to improve the SVM classification algorithm, the result will be compared with Graph Cuts approach that introduce spatial domain information of the result image of spectral classification with SVM. The proposed approach is tested on common scenes of urban imagery. The experimental results show satisfactory values and are very promising.

The separation of text / image is a major step in the processing of documents images. It consists of separating the document into two classes: text and image. In this context, it is important to implement approaches that can handle such documents. This paper presents a new method of separating text / image into a document image. The method developed is based on statistical analysis of texture and morphological operations. This is, initially, to calculate the energy for each pixel, using a method to extract features such as co-occurrence matrix and we have applied the morphological operations. we have tested this method on a database of 150 newspapers. The experimental results show satisfactory values and are very promising.

In this paper we present a Markovian method of classification of the satellite images, this method is based on a minimization of the posterior energy by the ICM method (iterated conditionnal mode) with the introduction of constraints of the spatial context. The originality of our method is the variability over the iterations of a temperature factor like in the Simulated Annealing algorithm (SA), indeed, the algorithm refines classification by re-estimating the statistics of the classes according to the previous iteration and by giving more and more importance to the contextual information through the parameter of temperature T. The implemented method is tested in an application of cartography of the land cover. The results are satisfactory comparing to other no-contextual classification methods such as the ISODATA and Kmeans.

The visual representation of information is an interesting way of communication both in industrial and science. The viewing of interactive data gives the users a fun and immediate access to information they are looking for. In this context and with existing technologies that make Internet applications rich in content, we can ask our selves for the relevance of a software architecture dedicated to this type of media. We tried to design an architecture responding to our expectations, and its feasibility study by the realization of an application in the context of geographic mapping

We present an automatic system for off-line printed Amazigh handwritten characters recognition, based on an hybrid approach combining Hidden Markov Models (HMM) and the Hough transform. After preprocessing on the image of the character, the representative chain of the character is build from the Hough Transformation. This chain is translated into sequence of observations that is used for the learning phase, by the HMM. Finally, we use the Forword classifier to recognize the character. The experimental results show the robustness of the system.

We propose to elaborate the land cover map of the Agadir region from the processing of a remote sensing image Landsat ETM+ covering the region. The methodology consists on an unsupervised classification of the image. We start by a pre-processing step in order to surround the useful information. We implement an iterative and auto organizer method known under the name of ISODATA, and apply it on the image. Finally we make a post classification, and a land validation.