Le présent manuscrit présente de nouvelles techniques d'extraction des structures : du dialogue de groupe, d'une part; de textes argumentatifs, d'autre part. Déceler la structure de longs textes et de conversations est une étape cruciale afin de reconstruire leur signification sous-jacente. La difficulté de cette tâche est largement reconnue, sachant que le discours est une description de haut niveau du langage, et que le dialogue de groupe inclut de nombreux phénomènes linguistiques complexes. Historiquement, la représentation du discours a fortement évolué, partant de relations locales, formant des collections non-structurées, vers des arbres, puis des graphes contraints. Nos travaux utilisent ce dernier paradigme, via la Théorie de Représentation du Discours Segmenté. Notre recherche se base sur un corpus annoté de discussions en ligne en anglais, issues du jeu de société Les Colons de Catane. De par la nature stratégique des conversations, et la liberté que permet le format électronique des discussions, ces dialogues contiennent des Unités Discursives Complexes, des fils de discussion intriqués, parmi d'autres propriétés que la littérature actuelle sur l'analyse du discours ignore en général. Nous discutons de deux investigations liées à notre corpus. La première étend la définition de la contrainte de la frontière droite, une formalisation de certains principes de cohérence de la structure du discours, pour l'adapter au dialogue de groupe. La seconde fait la démonstration d'un processus d'extraction de données permettant à un joueur artificiel des Colons d'obtenir un avantage stratégique en déduisant les possessions de ses adversaires à partir de leurs négociations. Nous proposons de nouvelles méthodes d'analyse du dialogue, utilisant conjointement apprentissage automatisé, algorithmes de graphes et optimisation linéaire afin de produire des structures riches et expressives, avec une précision supérieure comparée aux efforts existants. Nous décrivons notre méthode d'analyse du discours par contraintes, d'abord sur des arbres en employant la construction d'un arbre couvrant maximal, puis sur des graphes orientés acycliques en utilisant la programmation linéaire par entiers avec une collection de contraintes originales. Nous appliquons enfin ces méthodes sur les structures de l'argumentation, avec un corpus de textes en anglais et en allemand, parallèlement annotés avec deux structures du discours et une argumentative. Nous comparons les trois couches d'annotation et expérimentons sur l'analyse de l'argumentation, obtenant de meilleurs résultats, relativement à des travaux similaires. / This work presents novel techniques for parsing the structures of multi-party dialogue and argumentative texts. Finding the structure of extended texts and conversations is a critical step towards the extraction of their underlying meaning. The task is notoriously hard, as discourse is a high-level description of language, and multi-party dialogue involves many complex linguistic phenomena. Historically, representation of discourse moved from local relationships, forming unstructured collections, towards trees, then constrained graphs. Our work uses the latter framework, through Segmented Discourse Representation Theory. We base our research on a annotated corpus of English chats from the board game The Settlers of Catan. Per the strategic nature of the conversation and the freedom of online chat, these dialogues exhibit complex discourse units, interwoven threads, among other features which are mostly overlooked by the current parsing literature. We discuss two corpus-related experiments. The first expands the definition of the Right Frontier Constraint, a formalization of discourse coherence principles, to adapt it to multi-party dialogue. The second demonstrates a data extraction process giving a strategic advantage to an artificial player of Settlers by inferring its opponents' assets from chat negotiations. We propose new methods to parse dialogue, using jointly machine learning, graph algorithms and linear optimization, to produce rich and expressive structures with greater accuracy than previous attempts. We describe our method of constrained discourse parsing, first on trees using the Maximum Spanning Tree algorithm, then on directed acyclic graphs using Integer Linear Programming with a number of original constraints. We finally apply these methods to argumentative structures, on a corpus of English and German texts, jointly annotated in two discourse representation frameworks and one argumentative. We compare the three annotation layers, and experiment on argumentative parsing, achieving better performance than similar works.

Undervisningen i naturvetenskap ska bland annat bidra till att utveckla elevers förmåga att hantera olika källor och resonera naturvetenskapligt. Med utgångspunkt i olika argument för och emot mjölk och/eller vegetabiliska alternativ som måltidsdryck undersöks i denna studie hur elever i årskurs 5 diskuterar en fråga med naturvetenskaplig grund. Eleverna lyssnade i en liten grupp till några texter om mjölk och/eller vegetabiliska alternativ och fick sedan välja mellan mjölk, havre- och mandeldryck, samt diskutera och motivera sitt val. Eleverna hanterade i sina diskussioner inte data systematiskt utan bortsåg från data som inte stödde deras initiala val. De var inte heller kritiska mot de texter som de lyssnat till. Däremot ägnade de mycket tid åt att diskutera giltighet och relevans hos varandras argument. För att elever ska kunna resonera kring en fråga på ett naturvetenskapligt vis, krävs undervisning som ger dem möjlighet att träna på dessa förmågor.

Philosophiae Doctor - PhD / The basic premise of this study was that when a learner is confronted with two contradictory explanations of the same phenomenon, there is cognitive dissonance in the learner as the learner tries to determine which of the two explanations is correct. An argumentation-based instructional intervention programme (ABIIP) was created for and used on and by the Grade 10 learners in order to attempt to ameliorate this cognitive conflict. The purpose of this study was to determine the relative impact of that intervention programme on Grade 10 learners’ conceptions of lightning and thunder. The programme was designed to help learners to develop argumentative skills and use the acquired skills to negotiate and harmonise divergent and conflicting explanations of the nature of lightning and thunder that are propounded by different worldviews (Science and indigenous knowledge).

Magister Educationis - MEd / Two grade 11 classes of two high schools in Cape Town were taught some concepts of sound by means of indigenous musical instruments. The purpose was to find out the relative effects (or none) of two instructional strategies. Toulmin (1958)’s Argumentation Pattern, Ogunniyi (1997)’s Contiguity Argumentation Theory and Reiner et al. (2000)’s Substance Schema formed the theoretical framework. A pre-post-test quasi-experimental design was employed and data collated using questionnaires, a sound conceptual test, argumentation worksheets, and classroom observation schedules. One teacher taught the experimental group using dialogical argumentation while another teacher taught the comparative group using lecturedemonstration method, coupled with the use of ICTs for duration of four weeks. Data were analysed using a mixed (quantitative and qualitative) methods approach. The findings revealed that many the learners held some scientifically valid conceptions of sound prior to formal instruction. However, the learners also held many scientifically invalid conceptions in relation to the speed of sound in air, sound propagation, and sound produced by stringed instruments. The alternative conceptions of learners in the C group remained largely unchanged after instruction, while those of the E group changed appreciably, but not completely. The E group learners changed the alternative conceptions that were worked into structured argumentation activities better than those which were not. Also, the learners in both groups seemed to hold indigenous beliefs in relation to sound which did not seem to change after instruction. Most learners had a positive attitude towards the use of indigenous knowledge in the science class. No significant difference was found between male and female learners with respect to conceptual understanding of sound, indigenous beliefs, and interest in the integration of science and indigenous knowledge.

Magister Educationis - MEd / The study employed a quasi-experimental design to determine the effect of DAIM on learners’ conceptions of lightning. The experimental group was taught using DAIM while the control group was taught the same content using TLM. Data was collected using the Science Attitude Questionnaire (SAQ), Beliefs about Lightning Questionnaire (BALQ), Conceptions of Lightning Questionnaire (COLQ) and Science Achievement Test on Lightning (SATOL) which was used to determine learners’ overall performance on the topic of electrostatics. The data was analysed using both qualitative and quantitative methods. The findings of the study revealed that prior to the intervention (DAIM); the two groups of learners had both the scientific and the indigenous knowledge about lightning. A majority of the learners believed that lightning is caused by witches and traditional doctors. After being exposed to the DAIM most of the learners in the experimental group were found to have changed to the more scientific explanation of cause of lightning and protective measures against lightning. Few learners in the control were classified as possessing an equipollent worldview in terms of the CAT after the post tests. Some learners’ conceptions about lightning wavered between the scientific and traditional worldviews. The Science Attitude Questionnaire showed that both groups of learners had a positive attitude towards science. The findings also suggested that the inclusion of indigenous knowledge in science lessons promoted active participation from the learners, reinforced the learning of science because it promoted conceptual development and scientific literacy. The learners in the study also supported the integration of the scientific and the traditional worldviews about lightning. After the instruction, the learners in both groups seemed to still hold indigenous beliefs in relation to lightning. The post-test results showed that the DAIM group seemed to have been able to link the concept of lightning with electrostatics when they related lightning storms to electric discharge. The experimental group was found to be more elaborate in their explanations of the scientific nature of lightning than the control group which was not exposed to DAIM

Doctor Educationis / The central concern of this study has been to determine the effectiveness or otherwise of a combined Dialogical Argumentation Instructional Method (DAIM) and the Assessment for Learning Model (DAAFLIM) strategies in enhancing Tertiary and Vocational Education and Training (TVET) students’ conceptions of selected scientific topics. The extant literature has shown that students often hold other worldviews or funds of knowledge, which might be in conflict with canonical school science. In light of this, DAAFLIM has been chosen for a number of reasons: (1) it has been shown to be effective for revealing students’ scientific and alternative worldviews; (2) it provides the learning environment that encourages students to express themselves freely, exchange views with others, reflect on what they have learned, and even to change their minds in the face of stronger arguments; (3) it is compatible with the Curriculum and Assessment Policy Statements (CAPS) curriculum which emphasizes that educators should integrate school science with students’ indigenous knowledge as a way to make the former more relevant to their sociocultural environment; (4) assists educators to plan instruction in accordance with the needs of multicultural science classroom; and (5) the combination of classroom discourses with continuous or formative assessment (as exemplified by DAAFLIM), instead of the usual terminal summative assessment, tends to mitigate the fears that students usually associate with assessment. Specifically, a group of TVET students i.e. the Experimental group (E-group) was exposed to DAAFLIM while the other group i.e. the Control group (C-group) was exposed to traditional instruction method (TIM).

Ce travail doctoral a pour objectif d’analyser la signification du terme pardon dans tous ses emplois en langue française contemporaine. Le cadre théorique de la sémantique argumentative et de la TBS nous a permis de schématiser et de comprendre non seulement les différences de signification dans les emplois du nom commun mais également de discerner le mode d’action de la locution dans toutes ses fonctions argumentatives.Après avoir mis de côté ce qui appartient à la construction encyclopédique de la notion du pardon, à savoir le caractère d’un acte souverain, l’analyse argumentative a permis de décrire toute la richesse de la signification du terme en tant qu’état, processus et achèvement, ainsi que son caractère fondamental de promesse. Elle nous a également amené à dissocier d’un côté l’acte d’excuse performatif dans les échanges réparateurs et, de l’autre, des fonctions argumentatives variées dans les emplois conversationnels et de politesse.Notre démarche a permis d’enrichir les apports du dictionnaire et de l’approche lexicologique. A l’issue de l’analyse d’une centaine d’exemples, nous avons pu rapprocher le terme pardon de celui de paix. / This doctoral work aims to analyze the meaning of the term forgiveness in all its uses in contemporary French.The theoretical framework of the argumentative semantics and the TBS allowed us to schematize and understand not only the differences of meaning in the uses of the common noun but also to discern the mode of action of the phrase in all its argumentative functions.After setting aside what belongs to the encyclopedic construction of the notion of forgiveness, namely the character of a sovereign act, the argumentative analysis has allowed to describe all the richness of the meaning of the term as a state, a process and a completion, as well as its fundamental promise. It has also led us to dissociate on the one hand the performative excuse act in repairs and, on the other hand, various argumentative functions in conversational and politeness jobs.Our approach has enriched the contributions of the dictionary and the lexicological approach. At the end of the analysis of a hundred examples, we were able to bring the term forgiveness closer to that of peace.

No description available.

Ce travail de recherche est consacré à la formalisation et à la réalisation d’un processus de diagnostic pluridisplinaire. La particularité d’un tel diagnostic résulte du fait qu’il nécessite de nombreux spécialistes, chacun ayant des connaissances sur leur domaine. Le problème principal réside dans les interconnexions entre les domaines. Ces interconnexions peuvent ou non être connues et influer sur le diagnostic. Dans ce manuscrit, nous proposons de réaliser un diagnostic pluridisciplinaire l’aide d’un système multi-agents. Les agents élaborent un diagnostic local à un domaine puis, fusionnent leurs diagnostics afin d’obtenir le diagnostic pluridisciplinaire. Dans ce but, nous proposons un cadre d’argumentation et une méthode de fusion des diagnostics. Ensemble, ces deux propositions forment le modèle ANDi. / Sharing opinions among different participants is a useful and common way to build a constructive argumentation in order to solve complex problems that require the confrontation of different discipline areas. In such settings, experts build different arguments in relation to their own discipline area, then share and confront them to the other experts’ opinions. In this report we present an argumentative framework ANDi based on a multi-agent approach and Bayesian networks. In this framework, the agents support the elaboration of a global diagnostic from local ones. Local diagnostics are resulting of argumentations between group of experts from the same discipline area. We illustrate the use of this argumentation framework on the domain of fault diagnosis.

Nous considérons dans cette thèse la complexité algorithmique de problèmes émanant de deux formalismes de raisonnement non-monotone: l'abduction et l'argumentation. Le premier est destiné à formaliser le processus de trouver des explications pour une manifestation observée, le second (et plus récent) offre un cadre théorique pour formaliser le processus de l'argumentation. Nous nous concentrons sur le problème d'existence d'une explication pour l'abduction et sur le problème d'existence d'un argument pour l'argumentation. Dans le cadre de la logique propositionnelle dans son ensemble ces problèmes sont considérés comme étant des tâches algorithmiques difficiles (ils sont souvent situés au deuxième niveau de l'hiérarchie polynomial). Notre but est d'une part de comprendre les sources de difficulté, et d'autre part d'identifier des fragments de la logique propositionnelle dans lequels ces problèmes sont résolubles efficacement. Pour cela nous considérons ces problèmes d'abduction et d'argumentation dans deux cadres bien-établis qui permettent des classifications de complexité : Le cadre de Post et celui de Schaefer. Dans le cadre de Post, des restrictions sont faites sur les connecteurs autorisés dans les formules utilisées. Dans le cadre de Schaefer, on considère les formules en forme normale conjonctive généralisée, les "clauses" sont alors des applications de relations booléennes à des variables et on restreint le type des relations autorisées. / In this thesis we consider the computational complexity of problems from two central formalisms of nonmonotonic reasoning: abduction and argumentation. The first one is designed to formalize the process of finding explanations for some observed manifestation, the second (and more recent) one gives a theoretical framework to formalize the process of argumentation. We focus on the explanation-existence problem for abduction and on the argument-existence problem for argumentation. Considered in full propositional logic these problems are believed to be computationally costly tasks (they are often situated at the second level of the polynomial hierarchy). With the purpose of understanding sources of hardness and of identifying tractable fragments of propositional logic we consider several abduction and argumentation problems in two well-established settings allowing for complexity classifications. In the first one, Post's Framework, restrictions are made on the allowed connectives in the used formulae, whereas in the second one, Schaefer's Framework, one considers formulae in conjunctive normal form, where the clauses are generalized to applications of arbitrary Boolean relations to variables and one restricts the allowed type of relations. We discuss differences and common features between the explanation-existence and the argument-existence problem in function of the two chosen frameworks. Finally, we consider enumeration. In particular we consider the problem of enumerating all solutions (models) of a propositional formula by non-decreasing weight in Schaefer's framework (the weight of a model being the number of variables assigned to true).