Les Insoumis estiment que Kamala Harris a perdu parce que sa campagne n'etait pas assez radicale. Une grille de lecture qui permet de conforter la strategie de Melenchon.

Selon l'economiste Bertrand Martinot, il faut introduire une dose de capitalisation pour des raisons d'equite entre generations et d'efficience economique.

INTERVIEW. Pour le patron du groupe MoDem a l'Assemblee, il faut d'urgence sortir du bal des ego qui mine le << socle commun >> de Barnier et << etre utile aux Francais >>.

Apres la nomination d'Elon Musk a la tete d'un ministere de l'Efficacite gouvernementale, le ministre de la Fonction publique francais a exprime sa << hate >> de << partager les meilleures pratiques >>.

Le patron de l'ONG de defense des oceans Sea Shepherd avait fait une demande d'asile politique a Emmanuel Macron, au mois d'octobre.

Alors que les procureurs doivent requerir, ce mercredi 13 novembre a son proces, la presidente du groupe RN a l'Assemblee redoute, par-dessus tout, d'etre empechee de se presenter en 2027.

<< Incompetence >>, << radicalisme environnemental >>... L'Espagnole, proposee au poste de vice-presidente de la Commission en charge de la Transition ecologique, est bousculee par les deputes europeens.

L'infatigable Ridley Scott se repose pourtant sur ses lauriers avec cette suite inutile et empatee de son propre classique sorti en 2000. Spectaculaire mais vain.

LA CHRONIQUE DE WILLIAM THAY. Le nouveau gouvernement a provoque un climat anti-riches en rompant avec la politique fiscale attractive menee ces dernieres annees.

INTERVIEW. << J'aurais souhaite une contribution plus prospective et moins desobligeante a l'egard des barreaux de province >>, reagit M e Julie Couturier au rapport du P r Jamin.

¿Te gustaría escribir un libro? ¿Alguna vez lo has intentado? Sé que escribir un libro puede parecer un reto formidable. Muchas personas se preguntan si […]

Origen

This paper describes a new approach to document classification based on visual features alone. Text-based retrieval systems perform poorly on noisy text. We have conducted series of experiments using cosine distance as our similarity measure, selecting varying numbers local interest points per page, and varying numbers of nearest neighbour points in the similarity calculations. We have found that a distance-based measure of similarity outperforms a rank-based measure except when there are few interest points. We show that using visual features substantially outperforms textbased approaches for noisy text, giving average precision in the range 0.4-0.43 in several experiments retrieving scientific papers.

An OCR free word spotting method is developed and evaluated under a strong experimental protocol. Different feature sets are evaluated under the same experimental conditions. In addition, a tuning process in the document segmentation step is proposed which provides a significant reduction in terms of processing time. For this purpose, a complete OCR-free method for word spotting in printed documents was implemented, and a document database containing document images and their corresponding ground truth text files was created. A strong experimental protocol based on 800 document images allows us to compare the results of the three feature sets used to represent the word image.

Due to both the widespread and multipurpose use of document images and the current availability of a high number of document images repositories, robust information retrieval mechanisms and systems have been increasingly demanded. This paper presents an approach to support the automatic generation of relationships among document images by exploiting Latent Semantic Indexing (LSI) and Optical Character Recognition (OCR). We developed the LinkDI (Linking of Document Images) service, which extracts and indexes document images content, computes its latent semantics, and defines relationships among images as hyperlinks. LinkDI was experimented with document images repositories, and its performance was evaluated by comparing the quality of the relationships created among textual documents as well as among their respective document images. Considering those same document images, we ran further experiments in order to compare the performance of LinkDI when it exploits or not the LSI technique. Experimental results showed that LSI can mitigate the effects of usual OCR misrecognition, which reinforces the feasibility of LinkDI relating OCR output with high degradation.

This work describes an online handwritten character recognition system working in combination with an offline recognition system. The online input data is also converted into an offline image, and in parallel recognized by both online and offline strategies. Features are proposed for offline recognition and a disambiguation step is employed in the offline system for the samples for which the confidence level of the classier is low. The outputs are then combined probabilistically resulting in a classier out-performing both individual systems. Experiments are performed for Kannada, a South Indian Language, over a database of 295 classes. The accuracy of the online recognizer improves by 11% when the combination with offline system is used.

Neural network learning approach for color image restoration has been discussed in this paper and one of the possible solutions for restoring images has been presented. Here neural network weights are considered as regularization parameter values instead of explicitly specifying them. The weights are modified during the training through the supply of training set data. The desired response of the network is in the form of estimated value of the current pixel. This estimated value is used to modify the network weights such that the restored value produced by the network for a pixel is as close as to this desired response. One of the advantages of the proposed approach is that, once the neural network is trained, images can be restored without having prior information about the model of noise/blurring with which the image is corrupted.

In this paper, we propose a novel dexterous technique for fast and accurate recognition of online handwritten Kannada and Tamil characters. Based on the primary classifier output and prior knowledge, the best classifier is chosen from set of three classifiers for second stage classification. Prior knowledge is obtained through analysis of the confusion matrix of primary classifier which helped in identifying the multiple sets of confused characters. Further, studies were carried out to check the performance of secondary classifiers in disambiguating among the confusion sets. Using this technique we have achieved an average accuracy of 92.6% for Kannada characters on the MILE lab dataset and 90.2% for Tamil characters on the HP Labs dataset.

This paper systematically discusses the security and privacy concerns for e-learning systems. A five-layer architecture of e-learning system is proposed. The security and privacy concerns are addressed respectively for five layers. This paper further examines the relationship among the security and privacy policy, the available security and privacy technology, and the degree of e-learning privacy and security. The digital identity attributes are introduced to e-learning portable devices to enhance the security and privacy of e-learning systems. This will provide significant contributions to the knowledge of e-learning security and privacy research communities and will generate more research interests.

With the emergence of the Web 2.0 phenomena, technology-assisted social networking has become the norm. The potential of social software for collaborative learning purposes is clear, but as yet there is little evidence of realisation of the benefits. In this paper we consider Information and Communication Technology student attitudes to collaboration and via two case studies the extent to which they exploit the use of wikis for group collaboration. Even when directed to use a particular wiki designed for the type of project they are involved with, we found that groups utilized the wiki in different ways according to the affordances ascribed to the wiki. We propose that the integration of activity theory with an affordances perspective may lead to improved technology, specifically Web 2.0, assisted collaboration.

This paper presents a system that simulates, in real time, the volumetric deformation of muscles during human locomotion. We propose a two-layered motion model. The requirements of realism and real time computation lead to a hybrid locomotion system that uses a skeleton as first layer. The muscles, represented by an anatomical surface model, constitute the second layer, whose deformations are simulated with a finite element method (FEM). The FEM subsystem is fed by the torques and forces got from the locomotion system, through a line of action model, and takes into account the geometry and material properties of the muscles. High level parameters (like height, weight, physical constitution, step frequency, step length or speed) allow to customize the individuals and the locomotion and therefore, the deformation of the persons' muscles.

E-speranto is a formal language for generating multilingual texts on the World Wide Web. It is currently still under development. The vocabulary and grammar rules of E-speranto are based on Esperanto; the syntax of E-speranto, however, is based on XML (eXtensible Markup Language). The latter enables the integration of documents generated in E-speranto into web pages. When a user accesses a web page generated in E-speranto, the interpreter interprets the document into a chosen natural language, which enables the user to read the document in any arbitrary language supported by the interpreter.

The basic parts of the E-speranto interpreting system are the interpreters and information resources, which complies with the principle of separating the interpretation process from the data itself. The architecture of the E-speranto interpreter takes advantage of the resemblance between the languages belonging to the same linguistic group, which consequently results in a lower production cost of the interpreters for the same linguistic group.

We designed a proof-of-concept implementation for interpreting E-speranto in three Slavic languages: Slovenian, Serbian and Russian. These languages share many common features in addition to having a similar syntax and vocabulary. The content of the information resources (vocabulary, lexicon) was limited to the extent that was needed to interpret the test documents. The testing confirmed the applicability of our concept and also indicated the guidelines for future development of both the interpreters and E-speranto itself.

This work presents a first approach to the synthesis of Spanish Sign Language's (LSE) Classifier Constructions (CCs). All current attempts at the automatic synthesis of LSE simply create the animations corresponding to sequences of signs. This work, however, includes the synthesis of the LSE classification phenomena, defining more complex elements than simple signs, such as Classifier Predicates, Inflective CCs and Affixal classifiers. The intelligibility of our synthetic messages was evaluated by LSE natives, who reported a recognition rate of 93% correct answers.

This paper puts forward a verification method for compound purposes and compound reasons to be used during purpose limitation.

When it is absolutely necessary to collect privacy related information, it is essential that privacy enhancing technologies (PETs) protect access to data - in general accomplished by using the concept of purposes bound to data. Compound purposes and reasons are an enhancement of purposes used during purpose limitation and binding and are more expressive than purposes in their general form. Data users specify their access needs by making use of compound reasons which are defined in terms of (compound) purposes. Purposes are organised in a lattice with purposes near the greatest lower bound (GLB) considered weak (less specific) and purposes near the least upper bound (LUB) considered strong (most specific).

Access is granted based on the verification of the statement of intent (from the data user) against the compound purpose bound to the data; however, because purposes are in a lattice, the data user is not limited to a statement of intent that matches the purposes bound to the data exactly - the statement can be a true reflection of their intent with the data. Hence, the verification of compound reasons against compound purposes cannot be accomplished by current published verification algorithms.

Before presenting the verification method, compound purposes and reasons, as well as the structures used to represent them, and the operators that are used to define compounds is presented. Finally, some thoughts on implementation are provided.

Responsive open learning environments (ROLEs) are the next generation of personal learning environments (PLEs). While PLEs rely on the simple aggregation of existing content and services mainly using Web 2.0 technologies, ROLEs are transforming lifelong learning by introducing a new infrastructure on a global scale while dealing with existing learning management systems, institutions, and technologies. The requirements engineering process in highly populated test-beds is as important as the technology development. In this paper, we will describe first experiences deploying ROLEs at two higher learning institutions in very different cultural settings. The Shanghai Jiao Tong University in China and at the “Center for Learning and Knowledge Management and Department of Information Management in Mechanical Engineering” (ZLW/IMA) at RWTH Aachen University have exposed ROLEs to theirs students in already established courses. The results demonstrated to readiness of the technology for large-scale trials and the benefits for the students leading to new insights in the design of ROLEs also for more informal learning situations.

Siemens is well known for ambitious efforts in knowledge management, providing a series of innovative tools and applications within the intranet. References@BT is such a web-based application with currently more than 7,300 registered users from more than 70 countries. Its goal is to support the sharing of knowledge, experiences and best-practices globally within the Building Technologies division. Launched in 2005, References@BT features structured knowledge references, discussion forums, and a basic social networking service. In response to use demand, a new microblogging service, tightly integrated into References@BT, was implemented in March 2009. More than 500 authors have created around 2,600 microblog postings since then. Following a brief introduction into the community platform References@BT, we comprehensively describe the motivation, experiences and advantages for an organization in providing internal microblogging services. We provide detailed microblog usage statistics, analyzing the top ten users regarding postings and followers as well as the top ten topics. In doing so, we aim to shed light on microblogging usage and adoption within a globally distributed organization.

In this paper we present an approach to improving navigability of a hierarchically structured Web content. The approach is based on an integration of a tagging module and adoption of tag clouds as a navigational aid for such content. The main idea of this approach is to apply tagging for the purpose of a better highlighting of cross-references between information items across the hierarchy. Although in principle tag clouds have the potential to support efficient navigation in tagging systems, recent research identified a number of limitations. In particular, applying tag clouds within pragmatic limits of a typical user interface leads to poor navigational performance as tag clouds are vulnerable to a so-called pagination effect. In this paper, a solution to the pagination problem is discussed, implemented as a part of an Austrian online encyclopedia called Austria-Forum, and analyzed. In addition, a simulation-based evaluation of the new algorithm has been conducted. The first evaluation results are quite promising, as the efficient navigational properties are restored.

Collaborative Filtering(CF) is a well-known technique in recommender systems. CF exploits relationships between users and recommends items to the active user according to the ratings of his/her neighbors. CF suffers from the data sparsity problem, where users only rate a small set of items. That makes the computation of similarity between users imprecise and consequently reduces the accuracy of CF algorithms. In this article, we propose a clustering approach based on the social information of users to derive the recommendations. We study the application of this approach in two application scenarios: academic venue recommendation based on collaboration information and trust-based recommendation. Using the data from DBLP digital library and Epinion, the evaluation shows that our clustering technique based CF performs better than traditional CF algorithms.

One of the challenges in Dynamic Software Product Line (DSPL) is how to support the coordination of multiple instances of a service feature. In particular, there is a need for a decentralized decision-making capability that will be able to seamlessly integrate new instances of a service feature without an omniscient central controller. Because of the need for decentralization, we are investigating principles from self-organization in biological organisms. As an initial proof of concept, we have applied three bio-inspired techniques to a simple smart home scenario: quorum sensing based service activation, a firefly algorithm for synchronization, and a gossiping (epidemic) protocol for information dissemination. In this paper, we first explain why we selected those techniques using a set of motivating scenarios of a smart home and then describe our experiences in adopting them.

A feature model (FM) defines the valid combinations of features, whose combinations correspond to a program in a Software Product Line (SPL). FMs may evolve, for instance, during refactoring activities. Developers may use a catalog of refactorings as support. However, the catalog is incomplete in principle. Additionally, it is non-trivial to propose correct refactorings. To our knowledge, no previous analysis technique for FMs is used for checking properties of general FM refactorings (a transformation that can be applied to a number of FMs) containing a representative number of features. We propose an efficient encoding of FMs in the Alloy formal specification language. Based on this encoding, we show how the Alloy Analyzer tool, which performs analysis on Alloy models, can be used to automatically check whether encoded general and specific FM refactorings are correct. Our approach can analyze general transformations automatically to a significant scale in a few seconds. In order to evaluate the analysis performance of our encoding, we evaluated in automatically generated FMs ranging from 500 to 2,000 features. Furthermore, we analyze the soundness of general transformations.

Using pre-existing software components (COTS) to develop software systems requires the composition and adaptation of the component interfaces to solve mismatch problems. These mismatches may appear at different interoperability levels (signature, behavioural, quality of service and semantic). In this article, we define an approach which supports composition and adaptation of software components based on model transformation by taking into account the four levels. Signature and behavioural levels are addressed by means of transition systems. Context-awareness and semanticbased techniques are used to tackle quality of service and semantic, respectively, but also both consider the signature level. We have implemented and validated our proposal for the design and application of realistic and complex systems. Here, we illustrate the need to support the variability of the adaptation process in a context-aware pervasive system through a real-world case study, where software components are implemented using Windows Workflow Foundation (WF). We apply our model transformation process to extract transition systems (CA-STS specifications) from WF components. These CA-STSs are used to tackle the composition and adaptation. Then, we generate a CASTS adaptor specification, which is transformed into its corresponding WF adaptor component with the purpose of interacting with all the WF components of the system, thereby avoiding mismatch problems.

Feature modeling is an approach to represent commonalities and variabilities among products of a product line. Context-aware applications use context information to provide relevant services and information for their users. One of the challenges to build a context-aware product line is to develop mechanisms to incorporate context information and adaptation knowledge in a feature model. This paper presents UbiFEX, an approach to support feature analysis for context-aware software product lines, which incorporates a modeling notation and a mechanism to verify the consistency of product configuration regarding context variations. Moreover, an experimental study was performed as a preliminary evaluation, and a prototype was developed to enable the application of the proposed approach.

Visualization of search results is an essential step in the textual Information Retrieval (IR) process. Indeed, Information Retrieval Interfaces (IRIs) are used as a link between users and IR systems, a simple example being the ranked list proposed by common search engines. Due to the importance that takes visualization of search results, many interfaces have been proposed in the last decade (which can be textual, 2D or 3D IRIs). Two kinds of evaluation methods have been developed: (1) various evaluation methods of these interfaces were proposed aiming at validating ergonomic and cognitive aspects; (2) various evaluation methods were applied on information retrieval systems (IRS) aiming at measuring their effectiveness. However, as far as we know, these two kinds of evaluation methods are disjoint. Indeed, considering a given IRI associated to a given IRS, what happens if we associate this IRI to another IRS not having the same effectiveness. In this context, we propose an IRI evaluation framework aimed at evaluating the suitability of any IRI to different IR scenarios. First of all, we define the notion of IR scenario as a combination of features related to users, IR tasks and IR systems. We have implemented the framework through a specific evaluation platform that enables performing IRI evaluations and that helps end-users (e.g. IRS developers or IRI designers) in choosing the most suitable IRI for a specific IR scenario.

We investigate ambiguity for symmetric difference nondeterministic finite automata. We show the existence of unambiguous, finitely ambiguous, polynomially ambiguous and exponentially ambiguous symmetric difference nondeterministic finite automata. We show that, for each of these classes, there is a family of n-state nondeterministic finite automata such that the smallest equivalent deterministic finite automata have O(2ⁿ) states.

One important subject associated with personal authentication capabilities is the analysis of handwritten signatures. Among the many known techniques, algorithms based on linguistic formalisms are also possible. However, such techniques require a number of algorithms for intelligent image analysis to be applied, allowing the development of new solutions in the field of personal authentication and building modern security systems based on the advanced recognition of such patterns. The article presents the approach based on the usage of syntactic methods for the static analysis of handwritten signatures. The graph linguistic formalisms applied, such as the IE graph and ETPL(k) grammar, are characterised by considerable descriptive strength and a polynomial membership problem of the syntactic analysis. For the purposes of representing the analysed handwritten signatures, new hierarchical (two-layer) HIE graph structures based on IE graphs have been defined. The two-layer graph description makes it possible to take into consideration both local and global features of the signature. The usage of attributed graphs enables the storage of additional semantic information describing the properties of individual signature strokes. The verification and recognition of a signature consists in analysing the affiliation of its graph description to the language describing the specimen database. Initial assessments display a precision of the method at a average level of under 75%.

E-mail spam is no more garbage but risk since it recently includes virus attachments and spyware agents which make the recipients' system ruined, therefore, there is an emerging need for spam detection. Many spam detection techniques based on machine learning techniques have been proposed. As the amount of spam has been increased tremendously using bulk mailing tools, spam detection techniques should counteract with it. To cope with this, parameters optimization and feature selection have been used to reduce processing overheads while guaranteeing high detection rates. However, previous approaches have not taken into account feature variable importance and optimal number of features. Moreover, to the best of our knowledge, there is no approach which uses both parameters optimization and feature selection together for spam detection. In this paper, we propose a spam detection model enabling both parameters optimization and optimal feature selection; we optimize two parameters of detection models using Random Forests (RF) so as to maximize the detection rates. We provide the variable importance of each feature so that it is easy to eliminate the irrelevant features. Furthermore, we decide an optimal number of selected features using two methods; (i) only one parameters optimization during overall feature selection and (ii) parameters optimization in every feature elimination phase. Finally, we evaluate our spam detection model with cost-sensitive measures to avoid misclassification of legitimate messages, since the cost of classifying a legitimate message as a spam far outweighs the cost of classifying a spam as a legitimate message. We perform experiments on Spambase dataset and show the feasibility of our approaches.

Patient information in health care systems mostly consists of textual data, and free text in particular makes up a significant amount of it. Information retrieval systems that concentrate on these text types have to deal with the different challenges these medical free texts pose to achieve an acceptable performance. This paper describes the evaluation of four different types of information retrieval strategies: keyword search, search performed by a medical domain expert, a semantic based information retrieval tool, and a purely statistical information retrieval method. The different methods are evaluated and compared with respect to its appliance in medical health care systems.

Retrieving information or discovering knowledge from a well organized data center in general is requested to be familiar with its schema, structure, and architecture, which against the inherent concept and characteristics of cloud environment. An effective approach to retrieve desired information or to extract useful knowledge is an important issue in the emerging information/knowledge cloud. In this paper, we propose an ontology-based agent generation framework for information retrieval in a flexible, transparent, and easy way on cloud environment. While user submitting a flat-text based request for retrieving information on a cloud environment, the request will be automatically deduced by a Reasoning Agent (RA) based on predefined ontology and reasoning rule, and then be translated to a Mobile Information Retrieving Agent Description File (MIRADF) that is formatted in a proposed Mobile Agent Description Language (MADF). A generating agent, named MIRA-GA, is also implemented to generate a MIRA according to the MIRADF. We also design and implement a prototype to integrate these agents and show an interesting example to demonstrate the feasibility of the architecture.

Project monitoring plays a crucial role in project management, which is a part of every stage of a project's life-cycle. Nevertheless, along with the increasing ratio of outsourcing in many companies' strategic plans, project monitoring has been challenged by geographically dispersed project teams and culturally diverse team members. Furthermore, because of the lack of a uniform standard, data exchange between various project monitoring software becomes an impossible mission. These factors together lead to the issue of ambiguity in project monitoring processes. Ontology is a form of knowledge representation with the purpose of disambiguation. Consequently, in this paper, we propose the framework of an ontology-based real-time project monitoring system (ORPSM), in order to, by means of ontologies, solve the ambiguity issue in project monitoring processes caused by multiple factors. The framework incorporates a series of ontologies for knowledge capture, storage, sharing and term disambiguation in project monitoring processes, and a series of metrics for assisting management of project organizations to better monitor projects. We propose to configure the ORPMS framework in a cloud environment, aiming at providing the project monitoring service to geographically distributed and dynamic project members with great flexibility, scalability and security. A case study is conducted on a prototype of the ORPMS in order to evaluate the framework.

Data warehouses integrate and aggregate data from various sources to support decision making within an enterprise. Usually, it is assumed that data are extracted from operational databases used by the enterprise. Cloud warehousing relaxes this view permitting data sources to be located anywhere on the world-wide web in a so-called "cloud", which is understood as a registry of services. Thus, we need a model of dataintensive web services, for which we adopt the view of the recently introduced model of abstract state services (AS2s). An AS2 combines a hidden database layer with an operation-equipped view layer, and thus provides an abstraction of web services that can be made available for use by other systems. In this paper we extend this model to an abstract model of clouds by means of an ontology for service description. The ontology can be specified using description logics, where the ABox contains the set of services, and the TBox can be queried to find suitable services. Consequently, AS2 composition can be used for cloud warehousing.

Avec la fin du communisme, nous devions assister à la fin de l’histoire mais les crises économiques en ont décidé autrement. Et ce d’autant plus que depuis, les problématiques de déclassement social, territorial et de questionnement « identitaires » n’ont...

C’était il y a 3 ans et pourtant le souvenir de cette soirée est toujours là, pesante et iréelle. Exceptionnellement nous étions chez nous ce vendredi, en famille nous regardions la télévision. Si je n’ai pas souvenir du programme, je me souviens encore...

Une étude récente dévoile que que près de la moitié des maires ne souhaitent pas se représenter lors des prochaines municipales de 2020. Incontestablement, le mandat de maire est probablement le plus difficile, il revêt tant d’aspects et ce quelque soit...

Si la question de l’hystérisation des réseaux sociaux ne date pas des Gilets Jaunes, nulle doute qu’elle revient sur le devant de la scène. Je ne parle même pas des théories du complot lié à ce mouvement. Pour ma part, les choses sont claires, il est...

Comme on devait s’y attendre, l’attentat de Strasbourg n’a pas manqué, s’il en était encore besoin, de charrier son lot de rumeurs complotistes. Un regard rapide sur les réseaux sociaux de toute nature suffit à se rendre compte de ce triste phénomène....

Petite nouveauté : j'ouvre mon blog afin de faire vivre le débat et certains auront remarqué le changement d'intitulé. Désormais, Il y aura un nouveau contributeur en la personne de Vincent Grenier qui est entrepreneur et membre du Conseil National du...

L'actualité aidant, une personne sur twitter a exhumé un vieux billet que j'avais écrit en Mai 2011. A l'époque, je parlais du Parti Socialiste... sans fausse modestie, j'avais vu plutôt juste... Aujourd'hui, on peut déjà parler d'un autre "parti" mais...

Mine de rien les Gilets jaunes, quoiqu'on en pense, on réussit à faire bouger les lignes sur des éléments qu'on nous disait impossible jusqu'à présent. Au delà de certaines applications qui peuvent être laissées au débat, en matière de pouvoir d'achat,...

L'évènement est là, l'écrivain maudit et adulé - qu'on mérite - sort son dernier ouvrage, son nouveau chef d'oeuvre. Houellebecq a ça de bien, c'est qu'il permettra une nouvelle hystérisation du débat littéraire. Génie pour les uns, faussaire pour les...

Billet par Vincent Grenier. Il faut regarder cet extrait vidéo. Cela se passe devant nos futurs polytechniciens. Laurent Alexandre (heureux startuper millionnaire et apôtre du transhumanisme) nous explique pourquoi "les gilets jaunes sont des êtres substituables",...