This study aims at examining in detail the linguistic characteristics of sentential idioms (‘idiomatische Sätze’, IS) in German, e.g. Das kannst du dir an den Hut stecken.; Du hast wohl Tomaten auf den Augen!; Da lachen ja die Hühner!. In theoretically and empirically investigating their syntactic, semantic and pragmatic features and productive potential, it is shown that IS can only be described properly by looking at the interfaces between syntax, semantics, and pragmatics, and by systematically investigating pattern formation. In this respect, the study goes far beyond traditional phraseological research. As a theoretical background, the study refers to research on phraseology, sentence types and sentence mood, linguistic evaluation, morphological productivity, and the frameworks of Construction Grammar and Relevance Theory. Empirical evidence comes from a large corpus of written German, a linguistic experiment on productivity, and introspective data. With respect to syntax, the study shows that IS are bound to one or a few different sentence types, varying according to the degree of restriction of their speech act potential. The complex semantics of IS is shown to consist not only of a literal and an idiomatic, but also of a third, mediating representational level. The idiomatic meaning of IS is described as evaluative meaning. Pragmatically, IS are dependent on special contextual environments to obtain a proper interpretation. A context model is developed which makes use of certain categories of evaluation. Moreover, a relevance-theoretic approach is sketched in order to explain which functional advantages IS might have compared to non-idiomatic utterances. With respect to productivity, ten different idiomatic patterns of construction are identified and described in detail. It is experimentally demonstrated that these patterns are productive to a greater or lesser degree dependent on their internal syntactic, lexical, semantic and pragmatic features.

Thesis (M.S.W.)--University of Hong Kong, 1987.

The aim of the thesis "Syntactic Means of Rhematisation in Contemporary French in Comparison to Czech" is to describe syntactic means of rhematisation in French and denote its parallel in Czech language. First part of the thesis deals with the theoretical resources concerning Functional sentence perspective mainly in Czech-written literature by the most important authors (F. Daneš, J. Firbas, P. Sgall and others) and it also describes French conceptions by B. Combettes and M. Rothenberg concerning thematisation and rhematisation in French. The second part, which is practical, shows the concrete use of means described. The phrases are quoted from the literary texts contained in language corpus InterCorp.

En France et en Iran, le droit de l’arbitrage est le fruit d’une longue maturation historique. L’arbitrage en Iran plonge ses racines dans l’Antiquité. En raison de la culture du règlement pacifique des différends qui dominait alors, l’arbitrage s’est imposé comme le mode privilégié des règlements des différends. Dans le domaine de l’arbitrage commercial international, l’Iran s’est inspiré de la loi-type de la CNUDCI de 1997. Ce pays a également adhéré en 2001 à la Convention de New York. Le point de convergence entre les deux ordres juridiques français et iranien : ils ont cherché à instaurer des règles plus favorables pour la reconnaissance et l’exécution des sentences arbitrales, chacun ayant adopté une démarche différente. La première partie de cette étude est consacrée à la réception de la sentence arbitrale dans les deux systèmes. Seule la sentence définitive peut faire l’objet du contrôle et de l’exécution. Plus concrètement, la qualification en sentence internationale ou étrangère déterminera les règles applicables à la reconnaissance et l’exécution de celle-ci. La sentence arbitrale identifiée et reconnue est susceptible d'être contrôlée par le juge qui déterminera le sort de la sentence. La deuxième partie de notre analyse décrit ce contrôle qui peut être direct à l’occasion du recours en annulation ou indirect lors de la demande de reconnaissance et d’exécution. Ce mécanisme conduit in fine à protéger les droits de la partie gagnante et à empêcher l’abus, par la partie condamnée, des voies de recours. / In France and Iran, the law of arbitration is the result of a long historical development. Arbitration in Iran finds its roots in the Antiquity. Because of a culture favoring the pacific settlement of disputes, which was prevailing at that period, arbitration imposed itself as the privileged mode in resolving disputes. In the field of international commercial arbitration, Iranian legal system has been inspired by the UNCITRAL Model Law in the promulgation of its LICA in 1997. This country has also adhered to the New York Convention on the Recognition and Enforcement of Foreign Arbitral Awards in 2001. The converging point between the Iranian and French legal systems is the following: they have tried to establish the rules favoring the recognition and enforcement of arbitral awards, any of them adopting a different approach. The First Part of this study is allocated to the reception of the arbitral award in the two legal systems. Solely the final award could be subject to control and enforcement. More concretely, the qualification of the arbitral award as international or foreign award will determine the applicable rules in the matters of recognition and enforcement. The arbitral award, identified and recognized, could possibly be controlled by the judge who will determine its fate. The Second Part of our analysis describes the said control that could be direct, on the occasion of the action for setting aside the award, or indirect, when requesting the recognition and enforcement of the award. This mechanism will lead in fine to protecting the winning party’s rights and avoiding any abuse of the means of recourse by the losing party.

Bien que la prison peine demeure en situation d’échec, en particulier pour satisfaire à son objectif de réinsertion, en France, il y est encore largement fait recours, y compris pour la petite et moyenne délinquance. Dans un premier temps, cette thèse propose d’analyser les raisons de cet échec. Les peines de milieu ouvert, quant à elles, ne constituent pas des peines de substitution efficaces à l’emprisonnement en raison de leur manque de crédibilité dû à leur faible caractère punitif. Plutôt que de condamner catégoriquement la peine privative de liberté pour mettre en avant les bienfaits des peines de milieu ouvert, il est important de s’interroger sur la manière de rendre la peine privative de liberté plus efficace en matière de réinsertion. Pour ce faire, nous allons nous inspirer du modèle de détention ouvert et réfléchir sur l’opportunité de créer une sanction d’une sévérité intermédiaire entre la prison du milieu fermé et les mesures de milieu ouvert, une peine d’emprisonnement fondée sur la confiance et exécutée au sein du milieu semi-ouvert. / Although prisons continue to fail at their aim to resocialise offenders, in France prison sentences are widely handed down for minor and serious offences. This thesis analyses the reasons for this failure. Sentences prescribed by the community are not efficient substitutions for imprisonment because of their poor punitive character. Instead of categorically condemning imprisonment and replacing it with probation, it is important to ask how to make imprisonment a more effective resocialisation mechanism. From this, we draw our inspiration from open prisons and reflect on the appropriateness of introducing an intermediate sentence between de rigueur imprisonment and probation, a prison sentence based on trust and executed in a semi-open environment.

The recent uprise of end-to-end speech translation models requires a new generation of parallel corpora, composed of a large amount of source language speech utterances aligned with their target language textual translations. We hereby show a pipeline and a set of methods to collect hundreds of hours of English audio-book recordings and align them with their Italian textual translations, using exclusively public domain resources gathered semi-automatically from the web. The pipeline consists in three main areas: text collection, bilingual text alignment, and forced alignment. For the text collection task, we show how to automatically find e-book titles in a target language by using machine translation, web information retrieval, and named entity recognition and translation techniques. For the bilingual text alignment task, we investigated three methods: the Gale–Church algorithm in conjunction with a small-size hand-crafted bilingual dictionary, the Gale–Church algorithm in conjunction with a bigger bilingual dictionary automatically inferred through statistical machine translation, and bilingual text alignment by computing the vector similarity of multilingual embeddings of concatenation of consecutive sentences. Our findings seem to indicate that the consecutive-sentence-embeddings similarity computation approach manages to improve the alignment of difficult sentences by indirectly performing sentence re-segmentation. For the forced alignment task, we give a theoretical overview of the preferred method depending on the properties of the text to be aligned with the audio, suggesting and using a TTS-DTW (text-to-speech and dynamic time warping) based approach in our pipeline. The result of our experiments is a publicly available multi-modal corpus composed of about 130 hours of English speech aligned with its Italian textual translation and split in 60561 triplets of English audio, English transcript, and Italian textual translation. We also post-processed the corpus so as to extract 40-MFCCs features from the audio segments and released them as a data-set.

La sanction des infractions délictuelles constitue le nœud gordien d’une politique pénale contemporaine largement désapprouvée. Privilégiée, la peine privative de liberté manifeste de multiples défaillances portant atteinte à sa crédibilité. Les peines d’emprisonnement inférieures ou égales à cinq ans exécutées principalement en maison d’arrêt sont les plus problématiques car nombreuses. Elles sont le plus souvent inadaptées à la délinquance contemporaine. Si des alternatives, déjà anciennes et d’autres plus récentes, existent, elles ne sont que peu prononcées alors même que leur utilité sociale en matière de réinsertion et de lutte contre la récidive ne paraît plus à prouver. La punition hors les murs de l’infraction délictuelle est une thématique politique et législative récurrente. Cependant, elle ne parvient pas à s’ancrer véritablement dans le paysage judiciaire français ; son enracinement étant notamment freiné par le défaut d’adhésion des magistrats en partie conditionné par des mécanismes procéduraux pourvoyeurs de peine privative de liberté. L’heure du changement de paradigme a pourtant sonné dès lors que la peine d’emprisonnement ne permet pas d’agir sur les causes de l’acte infractionnel et du passage à l’acte. Il est également temps de soulager une situation carcérale exsangue ne permettant plus à l’Administration pénitentiaire d’assurer convenablement ses missions de garde et de réinsertion. Albert Camus écrivait qu’une société se juge à l’état de ses prisons, la société française ne peut persister à épandre une telle incarnation de sa politique pénale. Aussi, au travers de mécanismes procéduraux perfectionnés, de l’influence d’exemples positifs du droit comparé, grâce au faire-valoir des ressources intrinsèques des services pénitentiaires d’insertion et de probation, à la rédaction nouvelle d’articles fondamentaux du Code pénal et au déploiement de moyens adéquats, la punition hors les murs des délits passibles de cinq ans d’emprisonnement devient accessible et tangible. / The penalty of the criminal breaches constitutes the Gordian knot of a widely disapproved contemporary penal policy. Privileged, the custodial sentence shows multiple failures striking a blow at its credibility. The lower prison sentences or five-year-old equals executed mainly in detention center are the most problematic because numerous. They are badly adapted for the contemporary crime. If alternatives, already older and others more recent, exist, they are only little pronounced even though their social utility regarding rehabilitation and regarding fight against the repeat offense does not any more appear to prove. The punishment outside the walls of the criminal breach is a recurring political and legislative theme. Nevertheless, it does not succeed in anchoring really in the French judicial landscape; its implanting being slowed down in particular by the defect of membership of the magistrates partially conditioned by procedural mechanisms suppliers of custodial sentence. The hour of the change of paradigm nevertheless rang since the prison sentence does not allow to act on the causes of the unlawful act and the acting out. It’s also necessary time to relieve a pale prison situation not allowing anymore the Prison Service to assure suitably its missions of guard and rehabilitation. Albert Camus wrote that a society judges itself in the state of its prisons, the French society cannot persist in spreading such an embodiment of its penal policy. So, through sophisticated procedural mechanisms, the influence of positive examples of the comparative law, thanks to the use of the intrinsic resources of the prison departments of insertion and probation, to the new writing of fundamental articles of the Penal code and to the deployment of adequate means, the punishment outside the walls of the offences punished for five years of detention becomes accessible and tangible.

No description available.

No description available.

An Application Programming Interface (API) provides developers with a high-level framework that abstracts the underlying implementation of services. Using an API reduces the time developers spent on implementation, and it encourages collaboration and innovation from third-party developers. Making an API public has a risk: developers might use it inappropriately. Most APIs have a policy that states which behaviors are considered fraudulent. Detecting applications that fraudulently use an API is a challenging problem: it is unfeasible to review all applications that make requests. API providers aim to implement an automatic tool that accurately detects suspicious applications from all the requesting applications. In this thesis, we study the possibility of using machine learning techniques to detect fraud in Web APIs. We experiment with supervised learning methods (random forests and gradient boosting), clustering methods such as Density-Based Spatial Clustering of Applications with Noise (DBSCAN), and ensemble methods that combine the predictions of supervised learning methods and clustering methods. The dataset available contains data gathered when a developer creates an application and data collected when the application starts making HTTP requests. We derive a meaningful representation from the most important textual fields of the dataset using Sentence-BERT (S-BERT). Furthermore, we experiment with the predictive importance of the S-BERT embeddings. The method that achieves the best performance in the test set is an ensemble method that combines the results from the gradient boosting classifier and DBSCAN. Furthermore, this method performs better when using the S-BERT embeddings of the textual data of the applications, achieving an f1-score of 0.9896. / Ett API (Application Program Interface) ger utvecklare ett högnivåramverk som abstraherar den underliggande implementationen av tjänster. Användning av ett API reducerar tiden utvecklare lägger på implementation, och uppmuntrar samarbete med och innovation av tredjeparts-utvecklare. Att göra ett API publikt har ett risk: utvecklare kan utnyttja den på olämpliga sätt. De flesta API:erna har ett policy som beskriver beteenden som räknas som bedrägliga. Upptäckandet av applikationer som använder ett API på ett bedrägligt sätt är ett icke-trivialt problem, det är omöjligt att undersöka alla applikationer som skickar begäran till API:et. API leverantörerna siktar ständigt på att skapa ett automatiskt verktyg för att exakt upptäcka applikationer misstänkta för bedrägeri. I denna avhandling undersöks möjligheten av användning av maskininlärning för att upptäcka bedrägeri i Web API. Vi experimenterar med övervakad inlärningsmetoder (random forests och gradient boosting), klustring metoder som Density-Based Spatial Clustering of Applications with Noise (DBSCAN) och ensemble metoder som kombinerar prediktionerna av övervakad inlärningsmetoder och klustring metoder. Det tillgängliga datasetet innehåller data samlat när en utvecklare skapar en applikation och när den börjar skicka HTTP begäran. Vi härleder en meningsfull representation från de viktigaste textfälten i datasetet med hjälp av Sentence-BERT (SBERT). Dessutom experimenterar vi med den prediktiva betydelsen av S-BERT-inbäddningarna. Metoden som uppfyller den bästa prestandan i testsetet är ett ensemble metod som kombinerade resultaten från gradient boosting klassificeraren och DBSCAN. Denna metod presterar även bättre vid användning av S-BERT-inbäddnignarna av applikationernas textdata och därav uppnår ett f1-score på 0.9896.