Massive Open Online Courses (MOOCs) have broadened educational access but often overlook the diverse needs of learners, especially those with disabilities. This paper introduces the RECMOOC4ALL conversational chatbot, a Generative AI-powered tool crafted to enhance personalization, engagement, and accessibility within online learning environments. Unlike general-purpose chatbots, it collects user preferences, accessibility needs, and interaction patterns to support other components of the RECMOOC4ALL system such as the recommendation system and accessibility checker in delivering inclusive learning experiences. Leveraging technologies such as Speech-to-Text (STT), Text-to-Speech (TTS), multilingual support, and advanced AI models like GPT-3 and Whisper, the system fosters adaptive learning experiences tailored to individual requirements. This paper highlights the transformative potential of Generative AI in mitigating accessibility challenges and outlines the system's technical architecture and capabilities. Preliminary findings regarding user satisfaction and interaction efficacy, detailed in subsequent sections, underscore the tool's potential to significantly improve engagement and learning outcomes for students with disabilities.

ABSTRACT Volunteer Cloud computing, like traditional Cloud computing, has gained significant importance due to its ability to harness resources from individual personal machines, contributed voluntarily by their owners. In this paradigm, personal machine resources are shared voluntarily by their owners. This approach is particularly valuable for handling data‐intensive computations and storing Big Data. However, managing the potential unavailability of volunteer machines is crucial. Additionally, in Volunteer Cloud environments, energy consumption is becoming increasingly significant alongside execution time and cost. However, optimizing energy consumption remains a critical challenge, particularly in environments with dynamic and unpredictable availability of volunteer resources. This paper investigates the problem of energy‐efficient services placement in Volunteer Cloud environments. Specifically, we address the challenge of optimally deploying and running independent applications to minimize energy consumption while ensuring that physical machines are not overloaded and avoiding redundant deployments on the same machine. We propose two heuristic strategies to tackle this problem: the Dynamic Shortest Path Strategy (D‐SPS‐VC) for dynamic service placement and the Static Shortest Path Strategy (S‐SPS‐VC) for static service placement. These strategies are designed to optimize energy efficiency by considering constraints such as machine availability, capacity, and application duplication. The main contribution of this study is the development of these heuristic strategies, which are validated through a series of experiments that demonstrate their effectiveness in reducing energy consumption in Volunteer Cloud environments.

This study proposes an intelligent educational decision support system that empowers instructional designers to evaluate online educational contents in order to improve their design and effectiveness. The key challenge in developing it lies in automating and objectifying the evaluation process. To address this, the study pursues two main objectives. The first one is to propose a Multicriteria Approach for Learning Experience Analysis (MALEA) on which we have based the evaluation through the learners’ traces. The second objective consists of proposing the Approach for Content Success Prediction (ACSP), which can be used to evaluate educational content before its deployment. ACSP combines logistic regression and MALEA. This combination helps to guard against the possible imprecision of human judgment affecting the decision-making process. A case study is conducted and proves that the system meets the objective sought and thus is retained for online educational content evaluation. Results are promising with high values of precision, accuracy, specificity, and sensitivity. Different perspectives are finally proposed.

In the dynamically evolving of Massive Open Online Courses (MOOCs), the imperative for real-time and the efficient evaluation of learner engagement has never been more pronounced. Traditional methodologies, while providing foundational insights, often fall short in terms of objectivity and immediacy. This paper introduces an innovative system that uses advanced computer vision and machine learning algorithms to dynamically detect and analyze learner emotions and engagement levels during MOOC sessions. Additionally, this system facilitates the identification of areas for improvement and supports the design of personalized and engaging learning experiences, particularly for learners with disabilities. Our findings reveal that this system not only monitors the duration and intensity of learner engagement but also actively identifies moments of peak engagement and discerns learning patterns. This information enables the personalization of educational paths to suit individual learning styles, significantly enhancing engagement and overall MOOC effectiveness. Powered by affective computing, this technology seeks to make a difference in the field of education technology, transforming MOOCs into personalized learning environments that match the specific interests, goals and needs of each user.

Massive Open Online Courses (MOOCs) providers deliver courses to all students in a largely uniform manner, regardless of their individual needs. This ‘massiveness’ characteristic presents a number of challenges, most of which are connected to the lack of accessibility, inclusiveness, and personalization factors, particularly for learners with impairments and the elderly. Specifically, bridging the gap between learners' skills and individual variance owing to the lack of personalization and digital accessibility compliance, still remains burdensome. For this reason, incorporating accessibility and Universal Design for Learning (UDL) into online learning in general and MOOCs in particular might help establishing flexible learning experiences for all, taking into consideration learner variability and educational requirements by design. This paper sheds light on UDL and digital accessibility considerations within MOOCs and summarizes a literature review on MOOCs accessibility, paving the way to present our ongoing research aiming at enhancing MOOC accessibility harnessing artificial intelligence.

While massive research has been conducted to see how remote learning and teaching is conduct during the COVID-19 pandemic, less focus has been paid on remote special education for students with disabilities. Therefore, it is still not clear how those students learned and what types of challenges they faced. To fill this gap, this study first collected data from the literature via a systematic literature review, and from both 51 teachers and 21 students with disabilities who were involved in this remote teaching and learning experiences via surveys. It then conducted bibliometric, content and thematic analysis to draw conclusions. The obtained findings highlighted that online and offline remote teaching methods from home were applied. Additionally, different learning assessment methods, such as mini-projects and simple quizzes were adopted by teachers to assess the gained knowledge of students remotely, but none of these methods relied on emerging technologies, such as big data and learning analytics. Finally, parents were a core actor to maintain remote learning from home for students with disabilities.

A new approach for solving triangular band linear systems is established in this study to balance the load and obtain a high degree of parallelism. Our investigation consists to attribute both adequate start time and processor to each task and eliminate the useless dependencies which are not used in the parallel solve stage. Thereby, processors execute in parallel their related tasks taking account of the considered precedence constraints. The theoretical lower bounds for parallel execution time and the number of processors required to carry out the task graph in the shortest time are determined. Experimentations are realized on a shared-memory multicore processor. The experimental results are fitted to the values derived from the determined mathematical formulas. The comparison of results obtained by our contribution with those from triangular systems resolution routine belonging to the library PLASMA, Parallel Linear Algebra Software for Multicore Architectures, confirms the efficiency of the proposed approach.

Several research studies have highlighted that the traditional method of identifying dyslexia within learners is time consuming, expensive and might not be effective as some people acquired the skills to hide their disability. Particularly, no tool or method was reported in the Arab region (22 countries) that could help identify dyslexia within Arab learners. Therefore, this paper presents a developed and validated educational game to implicitly identify the level of dyslexia within learners based on their game play traces. The game was played by twenty-six children within a private school for special education with the supervision of experts from a private center for learners with disabilities. The obtained results showed that the accuracy level of identifying learners with dyslexia with the use of the game is high. Additionally, the experts reported a favorable perception and high technology acceptance degree towards the game. The findings of this research could enhance the educational technology field by providing an educational game design for implicit identification of dyslexia level.

Promoting intrinsic motivation in classrooms is the central focus of many researchers. Particularly, several studies found that students' individual differences might affect their level of intrinsic motivation. Since personality is considered as one of the most important indicators of individual differences, this study investigates the effect of personality on students' level of intrinsic motivation in a gamified and self-determined online environment. The obtained results showed that only extraversion, conscientiousness and openness personality traits significantly affect students' level of intrinsic motivation to learn. These findings can be considered by designers of gamified learning environments to provide more game design elements in order to enhance students' intrinsic motivation based on their personalities.

Designing an educational game can be a difficult task for individuals who do not have appropriate background towards game developments. Several game genres exist in the literature which can make beginners confused about which are the appropriate or required game elements and functionalities for a particular game genre. At the same time, many studies highlighted the effectiveness of mind-mapping techniques in enhancing learning and thinking task. Therefore, this study deploys the powerful graphical representation of Unified Modeling Language (UML) class diagram to present a mind-mapping guide for people who are developing educational role-playing games. A total of 1 13 students used the developed mind-mapping guide in a public Tunisian University, and the obtained results showed that the guide can: (1) significantly enhance beginner students’ level of knowledge regarding developing role-playing educational games; and, (2) promote the creative thinking when it comes to imagining educational role-playing game scenarios. Also, these results showed that students have a favorable attitude towards using the mind-mapping guide to learn and design their games.

While gamification has gained considerable scholarly and practitioner attention, less concern is paid regarding individual differences in gamified learning environments. Personality in particular is considered as the source of individual differences even within students. Therefore, this study investigates how personality can affect students' perceived preferences for eight game elements in a gamified Modular Object-Oriented Dynamic Learning Environment (Moodle). One hundred and five students, from a public Tunisian university, used this gamified Moodle during two years to learn "Object Oriented Design Methodology", "Basic Software" and "Information Monitoring Methodology" courses. The obtained results showed that only extraversion, conscientiousness and openness personality traits may affect students' perceived preferences for particular game elements. These findings can be exploited by researchers and practitioners to design personalized and gamified learning environments based on personality.

Gamification in education is gaining an increasing attention, since it enhances the learner's engagement and motivation. While carrying a big promise, prior research offers little guidance regarding how learners with different personalities may react to game elements in a gamified learning environment. Therefore, this study investigates the relationship between the learner's personality to be distinguished on an introvert/extrovert dimension and the attitude of preferring to use game elements. In this context, a gamified online Moodle Learning Management System was developed where fifty seven learners used it to learn "Object Oriented Design Methodology" course. These learners then answered two questionnaires about preferences on game elements and personality. Data gathered from the instruments was analyzed using the statistical ChiSquare test. The obtained results highlighted that the learner's personality can affect the attitude of preferring to use a particular game element. These findings can be exploited by researchers and practitioners to provide personalized and gamified learning experiences based on personality. They also can be used by designers by providing design guidelines for targeting specific learners based on personality.

In distributed computing, data scheduling is becoming an important field of research with the emergence of Big Data. High level features provided by software data scheduler often rely on data management policies - possibly user-defined - such as fault tolerance, multi-protocol file transfer, reliable and multi-tenant storage, security and data privacy, locality-aware data distribution etc. Nowadays, to execute data-intensive applications, such advanced features become necessary, and this means that data and task schedulers are capable to cooperate closely. In this paper, we propose a data driven cooperative platform by combining two existing middleware: XtremWeb-HEP, as the task scheduler, and BitDew, as the data scheduler. Taking advantage of both middleware, our solution allows user to select the suitable data scheduling strategy as well as the adequate task granularity which provide the optimal data distribution. To evaluate the efficiency of our approach, we compare different strategies of scheduling tasks and data and prove the efficiency of the cooperation of data and task schedulers to execute data-intensive applications.

Social networking services specifically Facebook has recently shown exponential growth in its number of users, which attracted researchers to investigate its educational potentials. While various studies have investigated learners' attitudes towards incorporating Facebook in classrooms, limited attention has been paid to examine learners' level of knowledge and motivation involved in online learning using Facebook. Therefore, this study proposes a new designed Facebook learning tool with various applied design strategies to enhance the learning process in classrooms. Thirty five learners, from a public University in Tunisia, participated in a quasi-experiment to validate this designed Facebook learning tool. The findings highlight that this tool has had a positive impact on the learners' level of knowledge regarding learning "game development" course. Besides, learners found this tool useful, interesting and safe, resulting in better engagement and a competitive attitude.

Various research work have highlighted the importance of modeling learner's personality to provide a personalized computer based learning. In particular, questionnaire is the most used method to model personality which can be long and not motivating. This makes learners unwilling to take it. Therefore, this paper refers to Learning Analytics (LA) to implicitly model learners' personalities based on their traces generated during the learning-playing process. In this context, an LA system and an educational game were developed. Forty five participants (34 learners and 11 teachers) participated in an experiment to evaluate the accuracy level of the learners' modeling results and the teachers' satisfaction degree towards this LA system. The obtained results highlighted that the LA system has a high level of accuracy and a "good" agreement degree compared to the questionnaire paper. Besides, the teachers found the LA system easy to use, useful and they were willing to use it in the future.

Infrastructure-as-a-Service container-based virtualization technology is gaining significant interest in industry as an alternative platform for running distributed applications. With increasing scale of Cloud Computing architectures, faults are becoming a frequent occurrence. Checkpoint-Restart is a key method to survive to failures in this context. However, there is a need to reduce the amount of checkpointing data as the Cloud is based on the pay-as-you-go model. This paper addresses the issue of garbage collection in LXCloud-CR and contributes with a novel decentralized garbage collection component “GC-CR”. LXCloud-CR, a decentralized Checkpoint-Restart model, is able to take snapshots of Linux Container instances and it uses replication to increase snapshots availability. LXCloud-CR contains a versioning scheme for each replica. The disadvantage refers to snapshots availability issues with versioning as the number of useless files grows. GC-CR is a decentralized garbage collector (checkpoint deletion) component that attempts to identify and eliminate old snapshots versions from the system in order to free storage space. Large scale experiments on the Grid'5000 testbed demonstrate the benefits of our proposal. Obtained results validate our model and show significant reduction of storage space consumption.

The increasing interest in 3D virtual human avatars has led to an expansion in the scale of their applications. One of the most significant uses of such technology is providing appropriate educational services for deaf and hearing impaired learners who experience immense difficulties in acquiring literacy skills in spoken languages. In fact, a great deal of effort has been devoted to create animated characters for rendering stories, poems, scientific terms, or any written text into signed languages (SLs). In this paper, we aim to investigate the effect of integrating a 3D signing avatar in an educational game and examine how such technology can boost the learning experience of deaf gamers. Particularly, we will seek, through a pilot study, to investigate whether the use of a signing avatar has a positive effect on the acquisition of SignWriting notation, a written form of SL, for those learners.

Integrating e-learning personalization systems in cloud computing environment is a relatively recent trend in technology enhanced learning. Cloud computing is a new model that allows users to access applications through Internet. It is characterized by several advantages such as reducing the cost of development, high availability of information, scalability and so on. In literature, several works have taken into consideration these advantages for applying this model in learning systems. However, very little research is available that focuses on selecting and assembling the best personalization parameters according to the teachers' preferences. This paper proposes a solution based on the cloud for the composition of personalization parameters with minimal cost. Through this solution, teachers can easily access these parameters from anywhere and anytime.

Recently, educational games are gaining increasing attention from researchers and practitioners in the learning field. This is because they are fun, motivating and immersive. However, designing educational games can be difficult for people who do not have appropriate background towards game development. This is further complicated by the fact that there are many genres of games (role playing, action, puzzle, etc.) and each genre has its own game elements and functionalities. This makes the game design process confusing, especially for beginners. This paper deploys the powerful graphical representation of Unified Modeling Language class diagram to present a generic model of educational role playing games. This model uses knowledge representation (game elements, functionalities and features) to support designing educational role playing games. Fifty eight game developers participated in evaluating the model and they have revealed its high perceived usefulness while designing educational role playing games. In addition, they reported that they were willing to use this model again in the future in their context.

In this paper, the Permutation Flow Shop Problem (PFSP) with total completion time is studied. In previous papers, we proposed parallel algorithms to solve the PFSP based on load balancing Branch and Bound approach. In this work, we propose other alternatives based on Genetic Algorithms to solve this problem. However, using efficiently genetic algorithm (GA) involves a complex interaction among its parameters. Therefore, finding good solutions for NP-hard problems (PFSP) cannot be guaranteed without proper setting of genetic parameters (selection, crossover, mutation, etc.). In this context, a parametric study is presented to select the best setting of GA parameters to solve the PSFP. Our study is completed with an extensive computational experiment, and the results are compared with known works in the literature. The results indicate that the proposed settings of the GA are efficient in finding better quality solutions.

Using computer games that support learning and teaching can offer various benefits to learners. The use of learning games can consume the attention of learners and increase their motivation and engagement which can then lead to stimulate learning. In particular, the use of learning versions of existing computer games has been found to have success in improving learners' motivation, increasing their desire to learn and making learning more enjoyable. However, most of the research to date on educational games, in particular learning versions of existing computer games, focused only on learner with typical development. Rather less is known about the designing of educational games for learners with special needs. For that, this paper presents a new educational game for deaf learners, called learning version of memory match game for deaf learners. This learning game was specially designed to foster vocabulary acquisition for deaf learners, to engage them, to keep them motivated and to render the learning experience with more fun.

Many researchers proved the efficiency of using educational games in the learning field. Thanks to the new wave of mobile technology, these games became even possible to run on mobile devices. On the other hand, various learners found learning the "computer architecture" subject dry and hard using the tradition way in classrooms. This paper presents a new mobile educational game which aims to teach the "computer architecture" subject. In particular, the different learning activities implemented within this game allow learners to learn this subject in a fun and interesting way.

The learner model is a key component of any adaptive E-Learning system, including educational games, as it stores the information about the learner, which can then be used to provide personalized experience. Recently, there has been growing interest in creating learner models using educational games. This is due to the limitations of traditional E-Learning systems, since the learner interaction with the computer is rather restricted in such systems. Educational games, on the other hand, not only stimulate learners by increasing their motivation and engagement, but also facilitate a lot of learner interaction that could be observed to create learner models. This paper presents a survey of the field of learner modelling using educational games. In particular, it describes the main methods of learner modelling. This paper also lists several educational games that are suitable for experimentation. Compiling this information can be important to the researchers and developers working in this field, especially to new researchers.

Technological advances, including the use of virtual reality, have contributed significantly to improving and expanding learning opportunities for people with hearing impairments over the past few years. For example, 3D human characters or avatars have succeeded in becoming widely used by deaf learners as an effective way of providing information in sign language. Indeed, these digital signers can be generated from text, speech, live-signing and even from sign language notations. In this context, we present in this paper tuniSigner, an avatar-based system able to display and interpret sign language transcriptions automatically, in the well known SignWriting system, via the visual-gestural modality. The main objective of the present is to help the hearing impaired learners to read and write from a written form of their native language. Reading and writing in their signed language could be a good starting point for improving their academic performance.

Summary form only given. The Arab League Educational, Cultural and Scientific Organization (ALECSO) has, as its primary responsibility, the promotion and coordination of educational, cultural and scientific activities for Arab countries. In this perspective, a "Plan for the Development of Education in the Arab Countries" (2008-2018) is being implemented by ALECSO. The implementation plan aims at systematically developing the Arab educational system. One of its fundamental dimensions, to be taken into consideration for Educational Improvement, deals with the effective use of ICTs in education to enhance learning/teaching practices. In this talk, we will explore a set of ongoing strategic ICT projects aiming to contribute in the innovation and the development of the educational Arab system: Promoting Open Educational Resources (OERs) for K-12 education in Arab countries, Creating MOOCs for Arabic language learning for native and non-native speakers, Promoting cloud computing technologies for education and research in the Arab world, and the ALECSO Apps store for promoting educational mobile content and apps.

Learning vocabulary is a very important part of learning a new language. It is vitally important for beginners and children to learn the keywords needed to understand what they hear and read, and to express their own ideas. Despite this importance, vocabulary acquisition stills the hardest part of the language and faces the problem of low motivation of learners. One of the effective ways of raising motivation and increase learners' desire to learn is to use learning games. This paper describes a learning version of memory match game specifically designed for second language acquisition. The objective of this learning game is to render the learning experience with more fun, to engage learners and to keep them motivated.

Recently, many works have investigated the identification of learner's working memory capacity (WMC) from his/her behavior in learning systems. These works used a mechanism, called dichotomic node network, able to create representation of the learner's WMC taking into account different theories from the literature. However, this mechanism could only provide a binary representation of learner's WMC (High or Low), whereas the WMC is estimated by 7 (plus or minus two) items. This paper presents an alternative based on fuzzy logic for precisely estimate the WMC of the learner while using learning system or playing educational games. The proposed approach can improve the accuracy of learner model and then enable the fine grained recommendations that positively affect the learners' learning.

Desktop Grid is among the success stories during last years by using volunteers nodes participating into projects. Now, with the emergence of Cloud Computing, the questions become where to take resources? and how to coordinate the resources? Our assumption is that Desktop Grid will continue to survive if we are able to transform the old-fashioned client/server architecture to new web oriented architecture to deliver services on demand. This paper revisits and extends the coordination protocol of BonjourGrid, a decentralized desktop grid system, based on the Publish-Subscribe paradigm and including a new tier for data management. The new protocol is designed according to a formal modelling using colored Petri nets. The protocol is veried and proved by CPN-Tools and implemented with Redis, a polpular net technology. We conducted out experiments on the Grid'5000 testbed using 300 nodes. We analyze the Redis performance and we demonstrate that the extended version of BonjourGrid system is fully operational.

For deaf and hard hearing people, learning any spoken language is not a natural or automatic process, it is rather a long and intensive task. As a matter of fact, the important differences between signed and oral languages significantly affect the ability of these individuals to develop their literacy skills and thus their knowledge and potential. To address that issue, a number of systems for representing SLs in written form have been proposed. Most have become widely used among academics for linguistic research, while a few others have begun to be used as educational tools in some countries. Typically, these notations may inadvertently create confusion for novice readers because of their static nature and the special symbols that they use, so, displaying its contents in the form of video or animation of humanlike character would be beneficial to their users. In this context, we present in this paper a new approach for automatically interpreting a sign language notation using avatar technology. Sign Writing will be the focus of our work because it depicts the more suitable method that could satisfy the deaf needs than other notations and which is being taught formally in some schools.

Nowadays the importance of accessibility in eLearning is well recognized. Numerous systems with this consideration have taken this path attempting to realize the merge between eLearning and accessibility. These attempts did not achieve significant exploits presenting several drawbacks preventing an environment successfully overlapping accessibility and eLearning. This paper presents our vision in designing an environment supporting eLearning accessibility. In order to ensure offering content in formats tailored to individual users based on their personal preferences [12], which guarantees an eLearning environment fully supporting the accessible learning content exploitation.

In recent years, there has been a growing interest in computer learning games. This is due to the belief that games can offer various learning benefits to learners. Specially, learning games which focus on programming language are needed today given that programming is seen as an important subject for learners. So, developing games which enhance learners' motivation for the programming subject is the main issue of this work. This paper describes a learning game for higher education. This new game is called Instruction right place Game and it allows learners to benefit from the drag and drop technology to construct a program in an amusing way.

In this paper, we describe a fully automatic learner modeling approach in learning management systems, taking into account learners' educational preferences including learning styles. We propose a learner model with three components: the learner's profile, learner's knowledge, and learner's educational preferences. The learner's profile represents the learner's general information such as identification data, the learner's knowledge implies the learner's interests on visited learning objects, and the learner's educational preferences are composed of the learner's preferences among visited learning objects and his/her learning style. In the proposed approach, all learner model components are automatically detected, without requiring explicit feedback. Indeed, all the basic learners' information is inferred from the learners' online activities and usage data, based on web usage mining techniques and a literature-based approach for the automatic detection of learning styles in learning management systems. Once learner models are built, we apply a hierarchical multi-level model based collaborative filtering approach, in order to gather learners with similar preferences and interests in the same groups.

With data intensive applications it can be interesting to resort to a distributed storage to reach scalability and avoid data-intensive problems. Storing data permanently on computing nodes can be an interesting approach especially with the frequent use and the large volume of this data. Moreover, processing large data is a computing intensive task which encourages parallel execution. Nevertheless, data placement on computing nodes should be optimal to reach load balancing. In this work, we investigate scheduling heuristics towards the optimization of data distribution on the computing nodes. Motivated by its capacity to control perfectly the common operations associated with data management, we use BitDew: a desktop grid middleware designed for large scale data management. With BitDew, we build a Data-Driven Master/Worker Platform to carry out the distribution of Magick, the OCR application based on Dynamic Time Warping (DTW) algorithm. We evaluate the benefit of the implementation of studied scheduling heuristics to achieve load balancing with both homogeneous and heterogeneous environment. We present experimental results which demonstrate the efficiency of our aproach.

Teachers and course authors need to apply various instructional strategies in order to personalize their courses. It is not a trivial task and the time, the efforts and the learning objects needed for preparing personalized learning scenarios depend on the personalization strategy to be applied. This paper presents an approach for selecting appropriate personalization strategies according to the feasibility of generating personalized learning scenarios with minimal burden on the author. Two metrics are proposed for putting in order and selecting useful personalization strategies. The calculation of these metrics is automated based on the analyses of the Learning Object Metadata (LOM) standard according to the semantic relations between data elements and learners' characteristics.

Usually semantic metadata are introduced to support effective search of relevant learning objects from LOR (Learning Object Repositories). We propose to use them to assist authors of SCORM (Sharable Content Object Reference Model) like learning objects. In fact, when we design new learning objects by reuse of existing ones, there are risks to have weaknesses mainly due to a bad understanding of the true nature of the reused objects. To overcome those risks we have defined an automated authoring assistance approach. In this paper we focus especially on how we use semantic metadata to produce indicators about the content of learning objects, more precisely about their complexity, heterogeneity and imbalance from a semantic perspective. To attend this specific objective we introduce the notion of "semantic space" and "semantic component". Then we use them to compute metrics which are aggregated together to produce meaningful indicators.

E-learning has made considerable progress in the last decades allowing different kinds of adaptation to the student profile or the learning objectives. But few E-learning standards take accessibility into account, making learners with disabilities find difficulties taking full advantage of E-learning platforms. Besides, most E-learning platforms present tests in a traditional format regardless of their known limitations, more important for applicants with special needs. However, it could be possible to integrate adaptive and more accessible E-testing functionalities in the E-learning platforms. This paper discusses the process of developing such an adaptive testing approach, which could be integrated in the Moodle platform, by presenting some of its principles, architectural elements and algorithms.

When there are many possible personalization strategies, it is important to select the most appropriate of them according to the specificities of the course. In this context, rises the contribution of this work, which is the analysis of learning objects as well as the metadata describing them for the selection of appropriate personalization strategy.

Semantic metadata are not yet fully integrated in most learning metadata profiles. Moreover, when it is supported it is only used to improve the quality of results returned by search engines in LOR (learning object repositories) and LMS (Learning Management Systems). Within the framework of APOGE project, we want to demonstrate the usability of semantic metadata, when they are provided, during the authoring phase to assist authors to improve the consistency of new designed learning objects. The added value of our approach is maximized in the case of SCORM (Sharable Content Object Reference Model) like learning objects' authoring. Indeed, when we have learning objects designed by reuse of existing ones there is risks to have weaknesses mainly due to a bad understanding of the true nature of the reused objects. To overcome those risks we have defined an automated authoring assistance approach. In this paper we put the emphasis on two semantic based methods used in our approach: semantic consistency checking and learning object's semantic space analysis.

In recent years, two major developments have been made. In one hand, E-learning has advanced in a remarkable manner, empowering education and creating very sophisticated environments. In the other hand, more importance has been attached to accessibility as it allows people with special needs to access to the Internet and to navigate and take advantage of web pages and services more efficiently. However, little attention has been devoted to make people with various disabilities benefit from the e-learning platforms. In this context, we are working on modeling the various types of learners with disabilities according to their preferences and their needs so as to insure their access to the various learning platforms. Our project consists on studying the field of accessible e-learning in order to give disabled learners the opportunity to access to the e-learning platforms just like any other students. This work is based on the OMG's Model Driven Architecture.

Today's e-learning environments are still far from being accessible for people with disability. Furthermore, the availability of accessibility guidelines, the diversity of the e-learning platforms and evolution of assistive technologies do represent just a partial solution. In fact, adopting the accessibility since the design phases could provide a rational solution. Therefore, the application of Model Driven Architecture (MDA) with a user centred approach turns out the most appropriate. This paper presents a new approach based both on different level abstraction models and IMS Access for all specification in order to allow the generation of a learning experience in different accessibility setting.

Abstract This paper proposes a decentralized and fault‐tolerant software system for the purpose of managing Desktop Grid resources. Its main design principle is to eliminate the need for a centralized server, therefore to remove the single point of failure and bottleneck of existing Desktop Grids. Instead, each node can play alternatively the role of client or server. Our main contribution is to design the PastryGrid protocol (based on Pastry) for Desktop Grid in order to support a wider class of applications, especially the distributed application with precedence between tasks. Compared with a centralized system, we evaluate our approach over 205 machines executing 2500 tasks. The results we obtain show that our decentralized system outperforms XtremWeb‐CH which is configured as a master/slave, with respect to the turnaround time. Copyright © 2009 John Wiley & Sons, Ltd.

We present a new approach called authoring by reuse to assist author during the design of SCORM like learning objects. It is now easy to design new learning objects by reusing existing ones connected together with dedicated operators. But the results are rarely satisfactory mainly due to their lack of consistency. In this paper we propose an assistance approach (i) providing indicators about the new designed learning object organized on three dimensions (complexity, reusability and pedagogy), (ii) computing metadata entries based on the information available about the reused learning objects and (iii) applying a rule based checking to ensure consistency.

In order to automatically provide the most appropriate learning objects to e-learners, a special interest should be given to the process of building the learnerspsila models. First, we need to identify which relevant information to include in the learnerpsilas model, while taking into account various pedagogical considerations, and second how to accurately infer the learnerspsila preferences and characteristics from their online behavior and activities. This paper presents a preliminary study of the possibility of integrating educational preferences in the learnerpsilas model and detecting them automatically within e-learning systems.

While the rapidly increasing number of users and applications running on desktop grid (DG) systems does demonstrate its inherent potential, current DG implementations follow the traditional master-worker paradigm and DG middlewares do not cooperate. To extend the DG architecture, we propose a novel system, called BonjourGrid, capable of 1) creating, for each user, a specific execution environment in a decentralized fashion and 2) contrarily to classical DG, of orchestrating multiple and various instances of desktop grid middlewares. This will enable us to construct, on demand, specific execution environments (combinations of Xtrem Web, Condor, Boinc middlewares). BonjourGrid is a software which aims to link a discovery service based on publish/subscribe protocol with the upper layer of a desktop grid middleware bridging the gap to meta-grid. Our experimental evaluation proves that BonjourGrid is robust and able to orchestrate more than 400 instances of Xtrem Web middleware in a concurrent fashion on a 1000 host cluster. This experiment demonstrates the concept of BonjourGrid as well as its potential and shows that, comparing to a classical desktop grid with one central master, BonjourGrid suffers from an acceptable overhead that can be explained.

Institutional Desktop Grid systems are attractive for running distributed applications with significant computational requirements. While the rapid increasing number of users and applications running on such systems does demonstrate the potential of Desktop Grid and Institutional Desktop Grid, current implementations follow the old-fashioned master-worker paradigm. Obviously, vulnerability to failures and permanent administrative monitoring are the disadvantages of client-server architectures. To bypass this, we proposed a novel system, called BonjourGrid, able to orchestrate multiple instances of Institutional Desktop Grid middlewares, able to remove the risk of single-source bottleneck and failure, and able to guaranty the continuity of services in a distributed manner. In this paper, we use BonjourGrid protocol, which is based on publish/subscribe paradigm, to show how to adapt it to fulfill all the requirements of a decentralised job scheduler. An evaluation proves that BonjourGrid is able to manage more than 100 applications instanciated in a concurrent way on an Institutional Desktop Grid. Analysing the execution of 100 applications with 2110 tasks during 3 hours demonstrates the potential of BonjourGrid concept and shows that, comparing to a classical Desktop Grid with one central master, BonjourGrid gives an acceptable overhead that can be explained.

Grid computing enables sharing, selection, and aggregation of a wide variety of geographically distributed computational resources such as supercomputers, clusters, storage systems and data sources. A grid presents them as one unified resource for solving large-scale and data intensive computing applications. A middleware supports applications in distributed computing environments by providing services that enable the interconnectivity and interoperability of applications, systems and machines. Considering the great development of grid systems, aggregation of several national and international grids offer more resources,improve and optimize the results. However, given the diversity of architectures and middlewares, it is important to think about solutions for interoperability between systems. In this paper, we present an architecture for enabling interoperability between the Tunisian national grid GTRS, based on the Peer-to-Peer middleware XtremWeb-CH, and the grid infrastructure of EU funded project EUMEDGRID based on gLite.In particular, we describe the solution related with the job submission mechanism that allows for users of EUMEDGRID to send jobs to GTRS.A new concept of Super Worker Node is proposed in this work to reach the interoperability between the two grids of GTRS and EUMEDGRID.

Several researches have demonstrated the effectiveness of graphical languages in knowledge representation and the numerous advantages of using graphs in the context of learning. On the other hand, many researches in the e-learning domain have focused on the personalization of learning materials by using adaptive educational hypermedia. However, most adaptive educational hypermedia systems do not benefit from the power of graphics because they do not use an expressive graphical language for presenting educational hypermedia to the learners. Our approach fills the existing gap between current educational hypermedia languages with the learnerspsila requirement of a standard graphical language for the visualization of adaptive hypermedia and the navigation through it. In this way, new possibilities are added to adaptive hypermedia expressiveness such as the presentation of the relations between concepts and the organization of these concepts in compartment allowing the arrangement of their semantics.

In spite of the fastest growing of technologies of information and communication in the education and the development of very sophisticated environments to improve learning and education, there are a few tools dedicated to deaf education due to the difficulty to create content in sign language. In this context, we present in this paper an ICT environment we developed to aid deaf people improving their social integration and communication capabilities. Our environment is a specialized LCS that generates multimedia courses to teach and learn sign language. These courses can be used either by deaf pupils to learn (or e-learn) sign language or also by hearing people to be able to communicate with deaf people. This educational environment uses mainly a Web-based interpreter of sign language developed in our research laboratory and called Websign. It is a tool that permits to interpret automatically written texts in visual-gestured-spatial language using avatar technology.

The World Wide Web (WWW) is becoming one of the most preferred and widespread mediums of learning. Unfortunately, most of the current Web-based learning systems are still delivering the same educational resources in the same way to learners with different profiles. A number of past efforts have dealt with e-learning personalization, generally, relying on explicit information. In this paper, we aim to compute on-line automatic recommendations to an active learner based on his/her recent navigation history, as well as exploiting similarities and dissimilarities among user preferences and among the contents of the learning resources. First we start by mining learner profiles using Web usage mining techniques and content-based profiles using information retrieval techniques. Then, we use these profiles to compute relevant links to recommend for an active learner by applying a number of different recommendation strategies.

In this paper, we compare four libraries for efficiently running threads when the performance of a CPU cores are degraded. First, we are interested by 'brute performance' of the libraries when all the CPU resources are available and second, we would like to measure how the scheduling strategy impacts also the memory management in order to revisit, in the future, scheduling strategies when we artificially degrade the performance in advance. It is well known that work stealing, when done in an anarchic way, may lead to poor cache performance. It is also known that the migration of threads may induce penalties if they are too frequent. We study, at the processor level, the memory management in order to find trade-offs between active thread number that an application should start and the memory hierarchy. Our implementations, coded with the different libraries, were compared against a Pthread one where the threads are scheduled by the Linux kernel and not by a specific tool. Our experimental results indicate that scheduler may perfectly balance loads over cores but execution time is impacted in a negative way. We also put forward a relation between the L1 cache misses, the number of steals and the execution time that will allow to focus on specific points to improve 'work stealing' schedulers in the future.

In this paper, we present an approach for the personalization of learning scenarios integrating different personalization parameters. In addition to "classical" ones such us learner preference, learning style, learner motivation and learner background, we integrate another aspect of personalization which focuses on the pedagogical approach. In fact, the context of learning, as well as the personalized object (contents and activities) and the personalization parameters depends strongly on the pedagogical approach.

We study, in this work, the load balancing of sort algorithm executed on a two cluster grid. Our solution is based on data partitioning. We use mainly geometrical interpretations to find out the optimal partition that reduces both communication and computing times in an heterogeneous context

In this paper, we present an adaptive educational hypermedia (AEH) based on case based reasoning (CBR) approach and XML language. CBR may be very useful in e-learning technologies. It should be viewed as a medium to be used in conjunction with the majority shared information system. In this context, we think that a standard way of marking up cases will facilitate this use. We suggest XML as the likely candidate to provide such standard. XML is used in order to represent the domain model, the user model and the case in the CBR system. CBR is used to simplify the process of personalization by reusing previous experiences

The aim of the paper is to introduce techniques in order to optimize the parallel execution time of sorting on heterogeneous platforms (processors speeds are related by a constant factor). We develop a constant time technique for mastering processor load balancing and execution time in an heterogeneous environment. We develop an analytical model for the parallel execution time, sustained by preliminary experimental results in the case of a 2-processors systems. The computation of the solution is independent of the problem size. Consequently, there is no overhead regarding the sorting problem.

The paper considers the problem of parallel external sorting in the context of a form of heterogeneous clusters. We introduce two algorithms and we compare them to another one that we have previously developed. Since most common sort algorithms assume high-speed random access to all intermediate memory, they are unsuitable if the values to be sorted don't fit in main memory. This is the case for cluster computing platforms which are made of standard, cheap and scarce components. For that class of computing resources a good use of I/O operations compatible with the requirements of load balancing and computational complexity are the key to success. We explore three techniques and show how they can be deployed for clusters with processor performances related by a multiplicative factor. We validate the approaches in showing experimental results for the load balancing factor.