• Publicationsclose
• 2017-2021close
• Open Access close
• Part of book or chapter of book close
• Conference object close
• CH close
• Archive ouverte UNIGE close

In this study, we compare the output quality of two MT systems, a statistical (SMT) and a neural (NMT) engine, customised for Swiss Post's Language Service using the same training data. We focus on the point of view of professional translators and investigate how they perceive the differences between the MT output and a human reference (namely deletions, substitutions, insertions and word order). Our findings show that translators more frequently consider these differences to be errors in SMT than NMT, and that deletions are the most serious errors in both architectures. We also observe there to be less agreement on differences to be corrected in NMT than SMT, suggesting that errors are easier to identify in SMT. These findings confirm the ability of NMT to produce correct paraphrases, which could also explain why BLEU is often considered to be an inadequate metric to evaluate the performance of NMT systems.

At a time when markets are globalizing, second language acquisition is increasingly important in current society. More than a way to stand out, multilingualism has become a necessity, especially in multilingual countries such as Switzerland. Like many aspects of the day to day life, language learning is being revolutionized by technological progress. Today, CALL (Computer-assisted language learning) software allows users to acquire a new language at their own pace, without any constraints of time and place. Thanks to the great evolution of speech recognition technology, it is now possible to interact orally with computers. On this basis, the Faculty of Translation and Interpreting of Geneva University developed its own software called CALL-SLT, a conversation partner exclusively based on speech recognition. Since its usefulness as a learning coach has already been proven, we propose a new experiment aiming to discover if CALL-SLT can also play the role of an automatic evaluator of the L2 pronunciation. 17 French speakers were given access to the tool and took a test specially developed for this purpose. In this paper, we will present the method we used and summarize the results. While these are not perfect, they are very encouraging and suggest that an evaluation of pronunciation with speech recognition is indeed possible.

Dealing with context-sensitive information is a crucial aspect in the management of semantic web data. Despite the importance of this topic, there is so far no accepted consensus regarding the precise way of encoding and even more reasoning on contextual knowledge. In this paper, we introduce an approach to reason over contextual knowledge in RDF, while committing to the semantics of a contextual description logic. The lines of this paper are many folds. First, we present an extension of OWL 2 DL for contexts, that we call {OWL 2 DLC. It is a two-dimensional web ontology language with one dimension for contextualized object knowledge and one dimension for contexts. Second, we define an OWLC profile for contextual reasoning, similar to OWL 2 RL. And finally, we demonstrate that the model can be practically implemented using existing semantic web technologies, especially using SPIN rules.

L'actualité des cas d'infanticide nous invite à interroger l'origine historique des folies de la maternité.

This chapter analyses the norms and principles concerning the management of transboundary waters under international law, as well as the provisions of the Water Framework Directive on the same issue under European Union law. Then, the focus is moved to the hydrographical situation of Italy and the waters the country shares with neighbouring States. Even though, unlike other European countries, Italy does not share major watercourses or water bodies with other States, in at least a couple of cases (Lake Maggiore and Lake Lugano shared with Switzerland, and River Isonzo/Soča shared with Slovenia) joint management and control systems are in force, which pre-date the WFD by many years.

International audience

Nous décrivons un jeu vidéo pour l'apprentissage de la conjugaison française, conçu dans une optique d'intégration du contenu dans la mécanique de jeu. L'évaluation montre une augmentation de 4 points aux tests, sur 20.