No description available.

Le monde est de plus en plus informatisé. Il y a de plus en plus de logiciels en cours d'exécution partout, depuis les ordinateurs personnels aux serveurs de données, et à l'intérieur de la plupart des nouvelles inventions connectées telles que les montres ou les machines à laver intelligentes. Toutes ces technologies utilisent des applications logicielles pour effectuer les taches pour lesquelles elles sont conçus. Malheureusement, le nombre d'erreurs de logiciels croît avec le nombre d'applications logicielles.Dans cette thèse, nous ciblons spécifiquement deux problèmes:Problème n°1: Il ya un manque d'informations de débogage pour les bugs liés à des exceptions.Cela entrave le processus de correction de bogues. Pour rendre la correction des bugs liées aux exceptions plus facile, nous allons proposer des techniques pour enrichir les informations de débogage.Ces techniques sont entièrement automatisées et fournissent des informations sur la cause et les possibilités de gestion des exceptions.Problème n ° 2: Il y a des exceptions inattendues lors de l'exécution pour lesquelles il n'y a pas de code pour gérer l'erreur.En d'autres termes, les mécanismes de résilience actuels contre les exceptions ne sont pas suffisamment efficaces. Nous proposons de nouvelles capacités de résilience qui gérent correctement les exceptions qui n'ont jamais été rencontrées avant. Nous présentons quatre contributions pour résoudre les deux problèmes présentés. / The world is day by day more computerized. There is more and more software running everywhere, from personal computers to data servers, and inside most of the new popularized inventions such as connected watches or intelligent washing machines. All of those technologies use software applications to perform the services they are designed for. Unfortunately, the number of software errors grows with the number of software applications. In isolation, software errors are often annoyances, perhaps costing one person a few hours of work when their accounting application crashes.Multiply this loss across millions of people and consider that even scientific progress is delayed or derailed by software error: in aggregate, these errors are now costly to society as a whole.We specifically target two problems:Problem #1: There is a lack of debug information for the bugs related to exceptions.This hinders the bug fixing process.To make bug fixing of exceptions easier, we will propose techniques to enrich the debug information.Those techniques are fully automated and provide information about the cause and the handling possibilities of exceptions.Problem #2: There are unexpected exceptions at runtime for which there is no error-handling code.In other words, the resilience mechanisms against exceptions in the currently existing (and running) applications is insufficient.We propose resilience capabilities which correctly handle exceptions that were never foreseen at specification time neither encountered during development or testing. In this thesis, we aim at finding solutions to those problems. We present four contributions to address the two presented problems.

This dissertation investigates the argument structure of verb phrases by identifying the syntactic roles and locations of the functional heads it consists of. Since the early 1990s, it has been widely accepted that the basic verb phrase consists of two distinct projections of a functional layer v/VoiceP, and a lexical layer √/VP (Chomsky 1995, Hale & Keyser 1993, Harley 1995; 2008a, Kratzer 1996, Marantz 1997). Recent developments in generative grammar, however, suggest that it may comprise of three projections (Pylkkanen 2002; 2008, Cuervo 2003, Collins 2005, Alexiadou et al. 2006, Harley 2013a, Merchant 2013): two functional projections – Voice, which introduces the external argument and licenses accusative Case; verbalizing v, which marks the eventuality type be/do/become/cause – and an acategorial lexical root (Cuervo 2003, Harley 2013a). In this dissertation, I explore the consequences of adopting the tripartite theory of verb phrases with two particular foci: (i) the structure of applicative and causative constructions and the interactions between the two; (ii) languages where the applicative and causative constructions are formed by attaching affixes to the verbal root. The main proposal of this dissertation is that various morpho-syntactic behaviors of applicatives and causatives and their cross-linguistic variation can be captured with two tools: (i) the hypothesis of the tripartite verb structure; and (ii) an understanding of the selectional criteria of the functional heads – Voice, Appl, and v – and their head-specific properties. The tripartite assumption solves for us some empirical puzzles and raises some new questions. I show that the three major achievements of the tripartite hypothesis are that it provides a syntactic account of the constraints on applicative and causative affix ordering, the distinct patterns of functional heads in their ability to introduce arguments, and the disparate morpho-syntactic behaviors of the three causative types due to the size of their complements. I then provide answers to some new questions that follow from the transition to the tripartite hypothesis. I elaborate the selectional mechanisms of the Voice, Appl, and v heads involved in applicatives and causatives. I reinterpret previously established facts about applicatives and causatives within the updated verbal structure.

Les travaux décrits dans ce mémoire portent sur la supervision du trac dans le c÷ur de réseau, au niveau de la couche applicative. Nous illustrons l'intérêt de la supervision dans la couche 7 grâce à trois études qui montrent les bénéces obtenus pour la sécurité et pour l'évaluation de modications d'architecture du réseau. La première étude utilise l'épidémiologie, la science qui étudie les causes et la propagation des maladies. L'épidémiologie fournit des concepts et des mé- thodes pour analyser à quels risques potentiels d'infection les PC des clients ADSL sont exposés. En particulier, nous voulons analyser les risques par rapport aux applications utilisées par les clients. Grâce à la supervision applicative du trac d'un large échantillon de clients ADSL dans le c÷ur de réseau, nous construisons un prol d'utilisation du réseau pour chaque client et nous détectons ceux qui génèrent du trac malveillant. A partir de ces données, nous étudions le lien entre certaines caractéristiques des clients avec leurs risques d'être infecté. Nous mettons en évidence deux applications et un syst ème d'exploitation qui constituent des facteurs de risque. Nous en déduisons un prol de client pour lequel existe un risque important de se voir infecter par un virus informatique. La deuxième étude porte sur l'opportunité pour un opérateur d'installer des caches P2P dans son réseau. Mettre les contenus P2P en cache pourrait être un bon moyen de réduire la charge dans le réseau. Cependant, les performances des caches sont aectées par de nombreux facteurs, en relation avec les caches eux-mêmes, mais également avec les propriétés de l'overlay, les contenus P2P et la localisation du cache. Dans le but d'évaluer l'utilité potentielle des caches P2P, nous eectuons la supervision à grande échelle du trac P2P, dans le réseau opérationnel de France Télécom. Après avoir étudié certaines propriétés du trac observé, nous simulons le fonctionnement d'un cache en nous basant sur les données recueillies pendant 10 mois. Nous sommes alors en mesure d'évaluer les performances d'un cache, en termes d'économie de bande passante si un cache était réellement déployé au moment de la capture de nos traces. De plus, nous étudions l'impact sur la performance de paramètres tels que les caractéristiques du cache et de mani ère plus importante le nombre de clients servis par le cache. Les résultats montrent que l'on pourrait réduire le trac P2P d'échange de chiers de 23% avec un cache passif. Enn, la troisème étude porte sur l'opportunité pour un opérateur de réseau de coopérer avec les réseaux P2P à travers une interface de type P4P. L'approche P4P permet aux clients P2P d'améliorer leur sélection de pairs dans l'overlay. Le trac P2P représente une proportion importante du volume du trac dans les réseaux. Cependant, les systèmes P2P les plus couramment utilisés à l'heure actuelle ne tiennent pas compte de l'infrastructure réseau sous-jacente. Grâce à la supervision applicative, nous déterminons les béné- ces de P4P, d'une part pour les applications P2P et d'autre part pour les opérateurs. Les résultats de cette expérimentation indiquent que les applications P2P ont besoin de plus d'informations que seulement l'AS de provenance des sources potentielles pour améliorer leurs performances. De plus nous montrons que le trac P2P inter-domaines pourrait être réduit d'au moins 72% grâce à P4P. Nous montrons donc dans ces travaux que la supervision applicative permet : d'analyser des phénomènes complexes liés à l'usage qui est fait du ré- seau, tels que la contamination par un ver ou un virus informatique ; d'évaluer, de manière précise et quantitative, l'impact de certaines modi cations d'architecture sur le trac opérationnel. D'une manière plus générale, nous illustrons l'importance du rôle de l'opé- rateur de réseau dans le déploiement et l'exploitation des services Internet, toujours plus gourmands en bande passante, que nous ne manquerons pas de voir apparaître à l'avenir. En eet la supervision applicative est un outil essentiel pour l'évaluation des protocoles et architectures mis en ÷uvre dans les services Internet, complémentaire des autres outils dans ce domaine.

This thesis shows that the mapping of semantics to syntax can be more complex than is generally assumed. In general, the mapping of semantics to syntax is thought to be many-to-one; for instance, many types of external argument roles are mapped to a subject position, and a theme or patient role is mapped to an object position. Contrary to this view, I show, by studying the syntax and semantics of external arguments, that one-to-one mapping between syntax and semantics is possible. External arguments are generally assumed to be introduced by a functional head, called Voice or v, regardless of the semantics of the argument, rather than being actual arguments of the verbs. A high Appl head similar to Voice has recently been argued to introduce external arguments as well as arguments of other semantic types. At present, no theories propose how these heads are distinguished in argument structure. This thesis articulates the differences between the external argument introducing heads and explores the consequences of these differences. Moreover, this thesis proposes a new type of event-related applicative, namely peripheral Appl. Like Voice and high Appl, peripheral Appl introduces an argument external to the verb phrase. The key differences among the external argument introducing heads are in their semantics as well as their syntactic position. Semantically, Voice is specified for agentivity, but high and peripheral Appls are specified for non-agentivity. Syntactically, high Appl merges below Voice, not above, while peripheral Appl can merge above Voice. An important result emerging from this thesis is that not all external arguments are treated in the same way in syntax: not only are agent and non-agent external argument roles mapped into different positions, but different types of non-agent roles are also mapped into different positions.

This thesis shows that the mapping of semantics to syntax can be more complex than is generally assumed. In general, the mapping of semantics to syntax is thought to be many-to-one; for instance, many types of external argument roles are mapped to a subject position, and a theme or patient role is mapped to an object position. Contrary to this view, I show, by studying the syntax and semantics of external arguments, that one-to-one mapping between syntax and semantics is possible. External arguments are generally assumed to be introduced by a functional head, called Voice or v, regardless of the semantics of the argument, rather than being actual arguments of the verbs. A high Appl head similar to Voice has recently been argued to introduce external arguments as well as arguments of other semantic types. At present, no theories propose how these heads are distinguished in argument structure. This thesis articulates the differences between the external argument introducing heads and explores the consequences of these differences. Moreover, this thesis proposes a new type of event-related applicative, namely peripheral Appl. Like Voice and high Appl, peripheral Appl introduces an argument external to the verb phrase. The key differences among the external argument introducing heads are in their semantics as well as their syntactic position. Semantically, Voice is specified for agentivity, but high and peripheral Appls are specified for non-agentivity. Syntactically, high Appl merges below Voice, not above, while peripheral Appl can merge above Voice. An important result emerging from this thesis is that not all external arguments are treated in the same way in syntax: not only are agent and non-agent external argument roles mapped into different positions, but different types of non-agent roles are also mapped into different positions.

This dissertation first addresses various shortcomings in definitions of “applicative” when compared to what is actually found across languages. It then proposes a four-way distinction among applicative constructions, relevant at least to Bantu, a large family of languages spoken in Sub-Saharan Africa. Because of the gradual nature of historical change, differences among construction types may be somewhat graded. In what are called Type A applicative constructions, the applicative morpheme expands the argument structure of its root by introducing an obligatorily present applied phrase. This expansion might result, but need not, in increased syntactic valence of the derived verb stem. Type A includes cases where the applicative on a lexicalized applicative stem still has the ability to introduce an applied phrase. In Type B applicative constructions, the applicative expands the argument structure of its root by introducing an obligatorily present applied phrase and performs other semantic/pragmatic functions on the applied phrase or on the whole clause (e.g. the applied phrase becomes the narrow-focused constituent in the clause). As in Type A, syntactic valence might be increased, but need not be. In Type C applicative constructions, the applicative does not introduce an applied phrase. Instead, it provides semantic nuances to the lexical meaning of its root (e.g. the action described by the root is performed to completion, repetitively, in excess, etc.). Unlike Type A and Type B, Type C constructions are not fully productive and may undergo lexicalization. Fourthly, in Pseudo-applicative constructions, the applicative morpheme found on a lexicalized stem does not introduce an applied phrase and does not perform semantic and/or pragmatic functions described for Type B and Type C. Because the last type, especially, has not been acknowledged in prior literature, the dissertation presents a historically informed case study of 78 pseudo-applicative forms in Tswana (S31), a southern Bantu language spoken in Botswana and South Africa. Finally, this study argues that both the synchronic functions of the Bantu applicative suffix *-ɪd and the lexicalization paths emerging from the study of Tswana pseudo-applicative forms support an original Location/Goal function of *-ɪd in Proto-Bantu, rather than an original Beneficiary function.

Le présent travail prend appui sur une synthèse de travaux déjà effectués dans l’équipe LaLIC (Langues,Linguistiques, Informatique, Cognition) ; il insiste sur l’articulation entre différents concepts, centrés sur la notion deréférentiel. Ma thèse fait partie d’une chaîne où sont approfondis certains concepts rendus de plus en plus opératoires.Mon travail prend appui sur les travaux sur le temps et l’aspect d’E. Benveniste, A. Culioli et J.-P. Desclés, enparticulier, il s’inscrit le développement de la théorisation du temps et de l’aspect entreprise par J.-P. Desclés et Z.Guentchéva. Dans cette approche, il existe cinq classes de concepts importants :1) Trois notions aspectuelles de base constitutives d’un procès : état, processus, événement ;2) Aspectualisation d’un procès sur un intervalle topologique d’instants ;3) Énonciation conceptualisée comme processus inaccompli ;4) Repérage par identification, différenciation et rupture ;5) Prise en compte de différents référentiels.Certains concepts avaient déjà reçus une formalisation rigoureuse. Au fil des analyses de plus en plus diversifiées, lanotion de référentiel est devenue de plus en plus importante dans le dispositif théorique et dans les descriptions associées.Il m’a donc été demandé de mieux préciser cette notion de référentiel en en proposant une approche formalisée quipermettrait ultérieurement des traitements informatiques. J’ai replacé cette notion dans le cadre théorique de latemporalité appréhendée par les langues en argumentant son introduction et en formalisant ses concepts. Afin de mieuxéprouver le caractère opératoire des concepts formalisés, il est apparu que leurs traductions dans un langage informatiquepermettraient de prouver leurs caractères réellement opératoires et de jeter des ponts solides orientés vers une utilisationopérationnelle des analyses linguistiques que nous proposons. / This dissertation support a synthesis of already accomplished works at the laboratory LaLIC(Languages, Linguistics, Informatics, Cognition) ; it emphasizes the articulation among different concepts, centered onthe notion of referential. My thesis forms a part of a chain where some rendered concepts are deepened more and moreoperatively. My work supports the works about the tense and the aspect of E. Benveniste, A. Culioli and J.-P. Desclés, inparticular, it is in line with the development of the theorization about the tense and the aspect proceeded by J.-P. Desclésand Z. Guentchéva. In this approach, there are five types of important concepts:1) Three basic aspectual notions constituent of a process (situation): state, process, event;2) Aspectualisation of a process (situation) on a topological interval of instants;3) Conceptualized enunciation as imperfective process;4) Reference by identification, differentiation and rupture;5) Recognition of different referentials.Some concepts are already received rigorous formalizations. Over the analyses being increasingly diversified, the notionof referential becomes more and more important in the theoretical plan and in the associated descriptions. It wasdemanded to me to improve the specification of this notion of referential while I proposed a formalized approach thatwould admit the computing treatments later. I put this notion within the theoretical framework of the temporality whichapprehended by the languages while arguing its introduction and while formalizing their concepts. To better perceive theoperative character of the formalized concepts, it was their translations in the computer language which enables to provetheir effectively operative characters and to dispose the solid link oriented towards an operational utilization of thelinguistic analyses that we propose.

Applying basic linguistic theory, a grammatical analysis of Gothic verbs formed with the previously underdescribed quasi-prefix (or preverb) miþ- shows that it forms what could be called an atypical type of applicative, though in two cases it seems instead to mark verbal congruence to an adverbial. Of the possible meanings of the etymologically identical preposition miþ, the comitative seems most salient for the quasi-prefix. An introduced core argument in these constructions do not behave like a typicaltransitive object, as it cannot be passivised, and so is likely rather an extension E (or indirect object) to the core. An underlying transitive object O remains as is, and can be passivised. The construction is only used when E pertains to the topic of the discourse, either an overlapping or a new topic. It is never a background comment, as is attested with peripheral arguments. The dative case of the introduced E seems to be semantically motivated, it being thecombination—comparable to a complex case marking—of the preposition and the quasi-prefix with the dative case that signals the comitative meaning. This construction is formally unrestrained, being applied to intransitive, ambitransitive, and transitive verbs, either with or without vowel gradation. In most cases the introduced E has a human referent, but it can also be a concrete object or an abstract. It cannot be inferred from this investigation whether Gothic allowed more than one applicative construction to be applied to the same verb. It would seem there are indications that this construction might have been a relatively new and developing construction in Gothic. In the verbal template, miþ- occupies the outermost position, in itself an indication of late grammaticalization. The one extant exception to this position could indicate an uncertainty of the translator regarding as to which its proper place was. The two instances with both miþ- and a comitative adverbial suggest—unless an artifact of translation—a possible development from peripheral argument to an atypical E-introducing applicative via verbal congruence.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The evolution of techniques for the development of equipment capable of translating the results taken from a diverse and complex set of tasks carried out by the human body, has become an important item on the scientific communities agenda. Within the context of such equipment, the force plate has shown itself capable of being implemented for the task of measuring phenomena, which before could only be observed visually. This work has as its objective to describe and develop an extensometric, multiaxial force plate, for use as a biomedical assistance apparatus. Sensors for the detection of forces such as strain gauges were used and fixed in predetermined areas on the four transducers, which are an integral and fundamental part of the force plate apparatus. The cell design was completed using geometry capable of coupling with the minimum of interference the two different forces present within the structure. In this manner it is possible to quantify the forces in three orthogonal directions. The prototype developed can be utilized, for example as an evaluation tool for both gait and stability research. In parallel to the development of the force plate, equipment for carrying out the conditioning and digitalization of signals originating from the load cells was designed and constructed. The control of the acquisition processes as well as the processing of signals received from the force plate is carried out by the use of a specific applicative, which makes the quantitative values of the forces, strength and momentum available, allowing for the calculation of the parameters used for the respective biomedical evaluations, especially those dealing with the oscillation of the human body in an erect near static posture. Still further the applicative provides a confidence ellipse obtained from the stability evaluations carried out along with the coordinates of pressure center and standard stability indexes obtained instantaneously. / A evolução das técnicas para desenvolver equipamentos capazes de traduzir o resultado de diversas tarefas complexas exercidas pelo corpo humano, tornou-se evidente para a comunidade científica. Dentre esses equipamentos, a plataforma de força mostrou ser uma ferramenta capaz de ser empregada para a mensuração de fenômenos antes só observados. Este trabalho tem por objetivo descrever o projeto e o desenvolvimento de uma plataforma de força extensométrica, multiaxial, para uso em instrumentação biomédica. Sensores para detecção de esforços, do tipo strain gauge, foram utlizados e fixados em regiões previamente determinadas nos quatro transdutores que compõem o conjunto da plataforma. O projeto das células considera uma geometria capaz de desacoplar, com mínima interferência, dois esforços diferentes, presentes na mesma estrutura. Assim, pode-se quantizar esforços em três direções ortogonais. O protótipo desenvolvido pode ser utilizado como ferramenta para avaliações de marcha e estabilidade, por exemplo. Paralelamente ao desenvolvimento da plataforma, um equipamento para realizar o condicionamento e digitalização dos sinais provenientes das células de carga foi projetado e construído. O controle dos processos de aquisição e processamento dos sinais advindos da plataforma de força é realizado através de um aplicativo específico que disponibiliza os valores quantitativos dos esforços de força e momento, os quais permitem o cálculo de parâmetros relevantes às avaliações biomédicas, particularmente em avaliações a cerca da oscilação do corpo humano na postura ereta quasi-estática. O aplicativo fornece ainda as elipses de confiança obtidas a partir das avaliações de estabilidade realizadas, juntamente com as coordenadas do centro de pressão e índices de estabilidade padrão e instantâneos obtidos. / Mestre em Ciências