Ce travail de recherche a pour objectif d’expliquer comment des enfants d’âges différents acquièrent des phénomènes linguistiques supposés complexes en roumain langue maternelle. Dans cette recherche nous avons retenu les phénomènes complexes suivants :- les verbes complexes ;- les relatives explicatives et déterminatives ;- les prédications complexes à deux compléments locatifs et deux temporels ;- les prédications complexes par réduction de deux prédications simples et- les prédications à trois verbes opérateurs.Pour décrire le parcours acquisitionnel des enfants, nous avons décrit les procédures de reformulation que des enfants de 6, 8, et 10 ans mettent en place lorsqu’ils restituent une histoire venant de leur être lue. Après l’analyse des procédures de reformulation pour chaque tranche d’âge nous sommes arrivées aux conclusions suivantes :A 6 ans :- les enfants ont tendance à effacer ou à simplifier les verbes complexes. Ces verbes complexes sont remplacés par des verbes lexicalement plus simples et plus fréquents.- les enfants restituent souvent les énoncés sources par des reformulations partielles et lacunaires.- ils ont tendance à reformuler la suite principale-relative en ne gardant que la principale. A cet âge les enfants utilisent un lexique simple, plus courant ou moins contraint.A 8 ans :- les enfants diversifient de plus en plus leurs procédures de reformulation : répétition, paraphrase sémantique, fonctionnelle et restructuration. Toutefois, la procédure de reformulation majoritaire est la paraphrase définitoire analytique lacunaire ce qui confirme le fait que les verbes complexes constituent un phénomène linguistique complexe.- les enfants reformulent les relatives explicatives en un nombre nettement supérieur par rapport aux enfants de 6 ans. Ils diversifient beaucoup plus les procédures de reformulation paraphrastique. Certaines reformulations maintiennent le degré de complexité de l’énoncé source (synthétisation de deux informations en un mot). Toutefois, les enfants à cet âge ont encore des difficultés à reformuler une relative déterminative contenant une métaphore ou à reformuler une prédication complexe composée de deux prédications élémentaires.A 10 ans :- les enfants reformulent sans difficulté les énoncés qui contiennent des relatives explicatives. Ils produisent un nombre supérieur de reformulations par paraphrase sémantique par rapport aux plus jeunes. C’est à cet âge également qu’ils complexifient de plus en plus les énoncés sources, cas de reformulation définitoire analytique complète et restructurations des deux prédications en sont un exemple.A cet âge, les enfants n’ont pas encore acquis les énoncés qui ont un haut degré de complexité syntaxique et sémantique comme la relative déterminative avec une métaphore, ce qui confirme l’hypothèse de la complexité de ce phénomène. / The aim of this research is to explain how children of different ages acquire linguistic phenomena which are supposed to be complex in Romanian as a mother-tongue language. In this research, we selected the following complex phenomena:- Complex verbs;- Non-restrictive and restrictive relative clauses;- Complex predication with two locative and two temporal complements.- Complex predication by reduction of two simple predications and- Predication with three operator verbsTo describe the learning path of children, we described the reformulation procedures that children aged 6, 8 and 10 follow when they reproduce a story that has just been read to them. After analyzing the reformulation procedures for each age group, we came to the following conclusions:At 6 years:- children tend to erase or simplify complex verbs. These complex verbs are replaced by lexically simpler and more frequent verbs.- they often reproduce source phrases by partial and incomplete reformulations.- they tend to reformulate relative clauses by simple phrases with a single predicate or erase information of complex phrases (non-restrictive and restrictive relatives). At this age, children use simple, more frequent or less restrained vocabulary.At 8 years:- children increasingly diversify their reformulation procedures: repetition, semantic and functional paraphrasing and restructuring. However, the reformulation procedure which is widely used is the incomplete analytical definitional paraphrase which confirms the fact that complex verbs constitute a complex phenomenon.- At this age, the number of reformulated non-restrictive clauses is significantly high as compared to that of children at the age of 6 year. They diversify much more paraphrastic reformulation procedures. Some reformulations maintain the same degree of complexity as the source phrases (synthesizing information). However, children at this age still have difficulties reformulating restrictive relative clauses that contains a metaphor or reformulating a complex predication composed of two elementary predications.At 10 years:- children reformulate non-restrictive phrases without any difficulties. They produce a high number of reformulations by semantic paraphrasing as compared to younger children. It is also at this age that they increasingly complexify the source phrase (the case of complete analytical definitional reformulation and restructuring of the two predications are good examples).- At this age, children have not yet acquired phrases with high degree of syntactic and semantic complexity such as the restrictive relative clauses with a metaphor, which confirms the hypothesis that this phenomenon is complex.

Thèses (M.A.)--Université de Sherbrooke (Canada), 1997. / Titre de l'écran-titre (visionné le 20 juin 2006). Publié aussi en version papier.

This dissertation addresses the modeling and analysis of a strategic risk management problem via a novel decision tree optimization approach, as well as development of enhanced Reformulation-Linearization Technique (RLT)-based linear programming (LP) relaxations for solving nonconvex polynomial programming problems, through the generation of valid inequalities and reduced representations, along with the design and implementation of efficient algorithms. We first conduct a quantitative analysis for a strategic risk management problem that involves allocating certain available failure-mitigating and consequence-alleviating resources to reduce the failure probabilities of system safety components and subsequent losses, respectively, together with selecting optimal strategic decision alternatives, in order to minimize the risk or expected loss in the event of a hazardous occurrence. Using a novel decision tree optimization approach to represent the cascading sequences of probabilistic events as controlled by key decisions and investment alternatives, the problem is modeled as a nonconvex mixed-integer 0-1 factorable program. We develop a specialized branch-and-bound algorithm in which lower bounds are computed via tight linear relaxations of the original problem that are constructed by utilizing a polyhedral outer-approximation mechanism in concert with two alternative linearization schemes having different levels of tightness and complexity. We also suggest three alternative branching schemes, each of which is proven to guarantee convergence to a global optimum for the underlying problem. Extensive computational results and sensitivity analyses are presented to provide insights and to demonstrate the efficacy of the proposed algorithm. In particular, our methodology outperformed the commercial software BARON (Version 8.1.5), yielding a more robust performance along with an 89.9% savings in effort on average. Next, we enhance RLT-based LP relaxations for polynomial programming problems by developing two classes of valid inequalities: v-semidefinite cuts and bound-grid-factor constraints. The first of these uses concepts derived from semidefinite programming. Given an RLT relaxation, we impose positive semidefiniteness on suitable dyadic variable-product matrices, and correspondingly derive implied semidefinite cuts. In the case of polynomial programs, there are several possible variants for selecting such dyadic variable-product matrices for imposing positive semidefiniteness restrictions in order to derive implied valid inequalities, which leads to a new class of cutting planes that we call v-semidefinite cuts. We explore various strategies for generating such cuts within the context of an RLT-based branch-and-cut scheme, and exhibit their relative effectiveness towards tightening the RLT relaxations and solving the underlying polynomial programming problems, using a test-bed of randomly generated instances as well as standard problems from the literature. Our results demonstrate that these cutting planes achieve a significant tightening of the lower bound in contrast with using RLT as a stand-alone approach, thereby enabling an appreciable reduction in the overall computational effort, even in comparison with the commercial software BARON. Empirically, our proposed cut-enhanced algorithm reduced the computational effort required by the latter two approaches by 44% and 77%, respectively, over a test-bed of 60 polynomial programming problems. As a second cutting plane strategy, we introduce a new class of bound-grid-factor constraints that can be judiciously used to augment the basic RLT relaxations in order to improve the quality of lower bounds and enhance the performance of global branch-and-bound algorithms. Certain theoretical properties are established that shed light on the effect of these valid inequalities in driving the discrepancies between RLT variables and their associated nonlinear products to zero. To preserve computational expediency while promoting efficiency, we propose certain concurrent and sequential cut generation routines and various grid-factor selection rules. The results indicate a significant tightening of lower bounds, which yields an overall reduction in computational effort of 21% for solving a test-bed of 15 challenging polynomial programming problems to global optimality in comparison with the basic RLT procedure, and over a 100-fold speed-up in comparison with the commercial software BARON. Finally, we explore equivalent, reduced size RLT-based formulations for polynomial programming problems. Utilizing a basis partitioning scheme for an embedded linear equality subsystem, we show that a strict subset of RLT defining equalities imply the remaining ones. Applying this result, we derive significantly reduced RLT representations and develop certain coherent associated branching rules that assure convergence to a global optimum, along with static as well as dynamic basis selection strategies to implement the proposed procedure. In addition, we enhance the RLT relaxations with v-semidefinite cuts, which are empirically shown to further improve the relative performance of the reduced RLT method over the usual RLT approach. Computational results presented using a test-bed of 10 challenging polynomial programs to evaluate the different reduction strategies demonstrate that our superlative proposed approach achieved more than a four-fold improvement in computational effort in comparison with both the commercial software BARON and a recently developed open-source code, Couenne, for solving nonconvex mixed-integer nonlinear programming problems. Moreover, our approach robustly solved all the test cases to global optimality, whereas BARON and Couenne were jointly able to solve only a single instance to optimality within the set computational time limit, having an unresolved average optimality gap of 260% and 437%, respectively, for the other nine instances. This dissertation makes several broader contributions to the field of nonconvex optimization, including factorable, nonlinear mixed-integer programming problems. The proposed decision tree optimization framework can serve as a versatile management tool in the arenas of homeland security and health-care. Furthermore, we have advanced the frontier for tackling formidable nonconvex polynomial programming problems that arise in emerging fields such as signal processing, biomedical engineering, materials science, and risk management. An open-source software using the proposed reduced RLT representations, semidefinite cuts, bound-grid-factor constraints, and range reduction strategies, is currently under preparation. In addition, the different classes of challenging polynomial programming test problems that are utilized in the computational studies conducted in this dissertation have been made available for other researchers via the Web-page http://filebox.vt.edu/users/dalkiran/website/. It is our hope and belief that the modeling and methodological contributions made in this dissertation will serve society in a broader context through the myriad of widespread applications they support. / Ph. D.

Ce travail de recherche se base sur l’hypothèse acquisitionnelle suivante : dans l’acquisition de leur langue maternelle les enfants produisent leurs propres énoncés à partir d’énoncés entendus préalablement en passant par des transformations appelées « reformulations ». Cette recherche comprend l’analyse des procédures de reformulation et de certains phénomènes complexes dans des productions orales en anglais langue maternelle (ALM) d’enfants de 8 et 10 ans et l’analyse des procédures de reformulation dans des productions orales en français langue étrangère (FLE) d’adolescents après environ 4 ans et demi d’apprentissage. A 8 ans en ALM, les enfants cherchent à simplifier les énoncés sources. A 10 ans, les enfants utilisent un plus grand nombre de procédures de reformulation différentes et ils utilisent plus de procédures de reformulation considérées comme complexes. Comme à 8 ans, les enfants de 10 ans ont toujours tendance à simplifier certains verbes complexes. Ceci montre également que ces verbes sont effectivement complexes. Enfin, les enfants reformulent plus souvent les relatives sources par des relatives que les enfants de 8 ans. Vers 17 ans et après environ 4 ans et demi d’apprentissage du FLE, la maîtrise de l’anglais langue maternelle jouerait un rôle déterminant dans l’acquisition du français langue étrangère : grâce à la proximité syntaxique de l’anglais et du français et à leur âge, les adolescents produisent plus de procédures de reformulation complexes. Les adolescents ont tendance à simplifier les énoncés sources quand l’énoncé source est complexe au niveau lexical comme les enfants en ALM. / This research is based on the new hypothesis on acquisition: in the course of the acquisition of their mother tongue, children produce their own utterances from previously heard utterances using transformations called “reformulations”. It consists of the analysis of the reformulation procedures and of some complex phenomena in children aged 8 and 10 years-old’s oral productions in English as a mother tongue (EMT) and of Analysis of the reformulation procedures in teenagers’ oral productions in French as a foreign language after approximately 4 to 5 years of learning.At 8 years-old in EMT, children tend to simplify the source utterances. At 10 years-old, children use a larger number of different kinds of reformulation procedures and they use more complex reformulation procedures. As at 8 years-old, 10-year-old children still tend to simplify some complex verbs. This also shows that these verbs are indeed complex. Lastly, children reformulate more often the source relative clauses with relative clauses than 8-year-old children. Towards 17 years-old and after 4 to 5 years of learning French as a foreign language (FFL), mastering the mother tongue would be a determinant factor in the acquisition of FFL: due to the syntactic proximity between the English and the French languages and to their age, teenagers produce paraphrases that are more complex. Teenagers tend to simplify source utterances when the source utterance is complex at a lexical and syntactical level, as do the children with English as a mother tongue.

Made available in DSpace on 2014-06-12T15:49:34Z (GMT). No. of bitstreams: 1 license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2009 / O processamento de consultas tem sido abordado como um problema central em ambientes dinâmicos e distribuídos. O ponto crítico do processamento, no entanto, é a reformulação da consulta submetida em um ponto origem em termos de um ponto destino, considerando as correspondências existentes entre eles. Abordagens tradicionais, em geral, realizam a reformulação utilizando correspondências de equivalência. Entretanto, nem sempre conceitos de um ponto origem têm correspondentes equivalentes no ponto destino, o que pode gerar uma reformulação vazia e, possivelmente, nenhuma resposta para o usuário. Neste caso, se o usuário considera interessante receber respostas relacionadas, mesmo que não precisas, é melhor gerar uma reformulação adaptada ou enriquecida e, por consequência, respostas aproximadas, do que nenhuma. Dentro deste escopo, o presente trabalho propõe um enfoque baseado em semântica, denominado SemRef, que visa integrar técnicas de enriquecimento e reformulação de consultas de forma a prover usuários com um conjunto de respostas expandidas. Reformulações exatas e enriquecidas são produzidas para permitir alcançar esse conjunto. Para tal, usamos semântica obtida principalmente de um conjunto de correspondências semânticas que estendem as normalmente encontradas na literatura. Exemplos de correspondências não usuais são closeness e disjointness. Além disso, usamos o contexto do usuário, da consulta e do ambiente como meio de favorecer o processo de reformulação e lidar com informações que somente são obtidas dinamicamente. Formalizamos as definições propostas através da Lógica Descritiva ALC e apresentamos o algoritmo que compõe o enfoque proposto, garantindo, através de propriedades aferidas, sua corretude e completude. Desenvolvemos o algoritmo SemRef através de um módulo de submissão e execução de consultas em um Sistema de gerenciamento de dados em ambiente P2P (PDMS). Mostramos exemplos que illustram o funcionamento e as vantagens do trabalho desenvolvido. Por fim, apresentamos a experimentação realizada com os resultados que foram obtidos

This dissertation is concerned with the development of algorithmic approaches for solving the minisum location and location-allocation problems in which the Euclidean metric is used to measure distances. To overcome the nondifferentiability difficulty associated with the Euclidean norm function, specialized solution procedures are developed for both the location and the location-allocation problems. For the multifacility location problem (EMFLP), two equivalent convex differentiable reformulations are proposed. The first of these is formulated directly in the primal space, and relationships between its Karush-Kuhn-Tucker (KKT) conditions and the necessary and sufficient optimality conditions for EMFLP are established in order to explore the use of standard convex differentiable nonlinear programming algorithms that are guaranteed to converge to KKT solutions. The second equivalent differentiable formulation is derived via a Lagrangian dual approach based on the optimum of a linear function over a unit ball (circle). For this dual approach, which recovers Francis and Cabot's (1972) dual problem, we also characterize the recovery of primal location decisions, hence settling an issue that has remained open since 1972. In another approach for solving EMFLP, conjugate or deflected subgradient based algorithms along with suitable line-search strategies are proposed. The subgradient deflection method considered is the Average Direction Strategy (ADS) imbedded within the Variable Target Value Method (VTVM). The generation of two types of subgradients that are employed in conjunction with ADS are investigated. The first type is a simple valid subgradient that assigns zero components corresponding to the nondifferentiable terms in the objective function. The second type expends more effort to derive a low-norm member of the subdifferential in order to enhance the prospect of obtaining a descent direction. Furthermore, a Newton-based line-search is also designed and implemented in order to enhance the convergence behavior of the developed algorithm. Various combinations of the above strategies are composed and evaluated on a set of test problems. Computational results for all the proposed algorithmic approaches are presented, using a set of test problems that include some standard problems from the literature. These results exhibit the relative advantages of employing the new proposed procedures. Finally, we study the capacitated Euclidean distance location-allocation problem. There exists no global optimization algorithm that has been developed and tested for this class of problems, aside from a total enumeration approach. We develop a branch-and-bound algorithm that implicitly/partially enumerates the vertices of the feasible region of the transportation constraints in order to determine a global optimum for this nonconvex problem. For deriving lower bounds on node subproblems, a specialized variant of the Reformulation-Linearization Technique (RLT) is suitably designed which transforms the representation of this nonconvex problem from the original defining space into a higher dimensional space associated with a lower bounding (largely linear) convex program. The maximum of the RLT relaxation based lower bound that is obtained via a deflected subgradient strategy applied to a Lagrangian dual formulation of this problem, and another readily computed lower bound in the projected location space is considered at each node of the branch-and-bound tree for fathoming purposes. In addition, certain cut-set inequalities in the allocation space, and objective function based cuts in the location space are generated to further tighten the lower bounding relaxation. Computational experience is provided on a set of randomly generated test problems to investigate both the RLT-based and the projected location- space lower bounding schemes. The results indicate that the proposed global optimization approach for this class of problem offers a promising viable solution procedure. In fact, for two instances available available in the in the literature, we report significantly improved solutions. The dissertation concludes with recommendations for further research for this challenging class of problems. Data for the collection of test problems is provided in the Appendix to facilitate further testing in this area. / Ph. D.

Skim, sweet buttermilk, and butter-derived aqueous phase components were used to re-emulsify low-melt butteroil into creams with a desired 20% milkfat. The implications of pasteurization process, homogenization sequence, and formulation on the physicochemical properties of reformulated and natural creams were investigated. Creams homogenized prior to pasteurization had significantly (p<0.05) greater amounts of milkfat surface material per gram of cream and per gram of lipid compared to creams homogenized after pasteurization. Significantly (p<0.05) higher percentages of available phospholipid also were associated with the milkfat surface material of creams homogenized prior to pasteurization. Phosphodiesterase, a marker enzyme within the native milkfat globule membrane, was in significantly (p<0.05) higher activity relative to protein on lipid globule surface when cream was homogenized prior to pasteurization. Creams that underwent pasteurization prior to homogenization had significantly (p<0.05) higher protein load associated with the milkfat surface material. Natural cream homogenized prior to pasteurization had significantly (p<0.05) greater milkfat surface material per gram lipid than natural cream homogenized after pasteurization and buttermilk / aqueous phase (AP) reformulated cream homogenized after pasteurization. In contrast, natural cream homogenized after pasteurization and BM / AP reformulated cream homogenized after pasteurization had significantly (p<0.05) greater amounts of protein per 10 mg of milkfat surface material than all other formulation / homogenization sequence combinations. In addition, natural cream homogenized prior to pasteurization, and skim milk (SM) reformulated cream pasteurized prior to or after homogenization had significantly (p<0.05) higher percentages of available phospholipid associated with the milkfat surface material than natural cream homogenized following pasteurization. Pasteurization temperature had a significant (p<0.05) effect on apparent viscosity of natural and reformulated creams. All UHT pasteurized natural and reformulated creams had significantly (p<0.05) greater apparent viscosities at all shear rates monitored than HTST pasteurized natural and reformulated creams. At a failing curve shear rate of 692 s-1 UHT pasteurized cream reformulated with BM / AP had significantly (p<0.05) greater viscosity than UHT pasteurized natural cream and cream reformulated with skim component. HTST pasteurized natural cream had significantly (p<0.05) greater viscosities than HTST pasteurized cream reformulated with skim component. Moreover, at a shear rate of 2769 s-1 and a rising shear rate of 1384 s-1 HTST pasteurized natural cream had significantly (p<0.05) greater viscosity than HTST pasteurized cream reformulated with buttermilk and aqueous phase. Creams formulated with skim component showed significant differences in creaming stability after 7 days of storage. On the other hand, natural cream and cream reformulated with buttermilk and aqueous phase showed significant differences in creaming stability after 9 days of storage. HTST pasteurized creams had greater creaming stability than UHT pasteurized creams. All creams feathered in a pH range of 5.09 to 5.31. Homogenization prior to UHT pasteurization resulted in creams rated "out of specification" because of poor sensory quality on day 1. Other processing sequences resulted in creams within sensory specifications. / Master of Science

L’objectif de notre thèse est de décrire les états linguistiques chez des enfants bilingues de 10 ans habitant à Paris, à travers leurs reformulations en français et en japonais de l’histoire « Tom et Julie ». Cette histoire comporte plusieurs éléments qui permettront d’analyser l’acquisition linguistique. Nous examinerons premièrement la « complexité » des reformulations. Deuxièmement, en examinant les procédures de reformulation, nous catégoriserons six formes de reformulations. Cette classification nous permettra de saisir les tendances des reformulations chez les enfants bilingues. Troisièmement, nous ferons l’étude comparée des reformulations en français et en japonais des deux textes. En dernier lieu, nous étudierons individuellement les reformulations en français et en japonais chez les enfants bilingues. Comment les enfants manipulent-ils les deux langues, qui sont si éloignées l’une de l’autre, pour exprimer la même histoire ? Quelle performance linguistique les enfants nous montrent-ils ? Peut-on trouver des règles linguistiques qui permettent aux enfants bilingues d’apprendre efficacement les deux langues ? Notre étude tente d’expliquer l’état linguistique des enfants bilingues en répondant à ces questions. / The goal of our thesis is to describe the linguistic states of bilingual children who are 10 years old and live in Paris, through their reformulation in French and in Japanese of the story « Tom et Julie ». This story contains some elements that permit us to analyze linguistic acquisition. Firstly, we examine the complexity of reformulations. Secondly, in examining the procedures of reformulation, we categorize six forms of reformulation. This classification will permit us to catch the tendencies of reformulations in the case of bilingual children. Thirdly, we compare French reformulation and Japanese reformulation in the two texts. Lastly, we will study these reformulations individually. How children manipulate French and Japanese that are so distant from each other, to express the same story ? What linguistic performance do they show ? Can we find linguistic rules which permit bilingual children to learn efficiently two languages ? Our study tries to explain the linguistic state of bilingual children in responding to these questions.

A tradução de Sérgio Milliet dos Essais de Michel de Montaigne para o português brasileiro envolve um trabalho de leitura, interpretação e escritura em várias frentes: estilística, retórica, linguística, antropológica, histórica, etc. O objetivo deste trabalho é identificar como foram traduzidos os pontos de inflexão nos quais os discursos do Outro aparecem no de Montaigne. O fio condutor do estudo pelo texto montaigneano representado pelo o paratexto páginas de rosto, prefácios e notas das edições francesas e das edições usadas na tradução leva, em torno da noção de reformulação, às heterogeneidades enunciativas e à tradução da escritura. O percurso traçado faz com que a reformulação, no sentido amplo da comunicação, seja pensada como uma noção operatória na abordagem de questões discursivas. O princípio que norteia o trabalho é baseado numa análise do discurso que considera o interdiscurso como origem de todo discurso e a heterogeneidade enunciativa como constitutiva da linguagem. Assim, a leitura/interpretação dos capítulos leva em conta a maneira como alguns traços da heterogeneidade a citação e a ironia se realizam na escritura do texto de Montaigne e nas traduções. A partir da observação empírica das traduções em francês moderno e em português brasileiro, nota-se que esses traços, diferentemente reformulados, trazem para dentro dos novos textos características que podem ser atribuídas às representações, sociais e individuais, dos próprios tradutores. / The Brazilian Portuguese translation of the Essais de Michel de Montaigne by Sérgio Milliet involves an effort of reading, interpretation and writing in many ways: stylistic, rhetoric, linguistic, anthropological, historical, etc. This work aims at the identification of how those points of inflexion in which the Others discourse shows up in Montaignes discourse were translated. The paratextual title pages, prefaces and notes of the French editions were the connecting thread through Montaignes texts to the enunciative heterogeneities and to the translation. The reformulation is, in the broad sense of communication, an operating notion in the approach of discourses questions. The research is based on the discourse analysis, which considers the interdiscourse as the origin of all discourse and the heterogeneity as a constructive element of language. The reading/interpretation of chapters takes into account the heterogeneities citation and irony that appear in Montaignes writing and in the translations. From the empirical observation of translations to modern French and to Brazilian Portuguese we noticed that these features, formulated in a different way, bring to the new texts some characteristics that may be imputed to the social and individual representations of the translators themselves.

Les travaux présentés dans cette thèse traitent des apports de l'algorithmique génétique à la conception de systèmes de recherche d'information adaptatifs aux besoins des utilisateurs.<br>Notre étude se focalise en premier lieu, sur l'analyse des différents modèles et stratégies de représentation et de recherche d'information. Nous mettons notamment en évidence, leur contribution à la résolution des problèmes inhérents à la recherche d'information. <br>En second lieu, notre intérêt s'est porté sur l'étude des algorithmes génétiques. Nous basant alors sur leur robustesse, théoriquement et expérimentalement prouvée, nous intégrons ces algorithmes à la mise en oeuvre de notre approche d'optimisation de requête.<br>Nous présentons une approche de recherche d'information qui intègre une stratégie de reformulation de requête par injection de pertinence, fondée sur l'hybridation d'un processus d'optimisation génétique, à un modèle de recherche de base. Nous proposons un algorithme spécifique à la recherche d'information, par l'intégration d'opérateurs génétiques augmentés par la connaissance du domaine d'une part, et d'une heuristique permettant de résoudre le problème de multimodalité de la pertinence d'autre part. L'heuristique de nichage en l'occurrence, est diffusée globalement lors de l'évolution de l'AG. La population est alors organisée en niches de requêtes effectuant une recherche parallèle et coopérative d'informations.<br>Nous évaluons enfin notre approche de recherche d'information, au travers d'expérimentations réalisées à l'aide du système Mercure, sur la collection de référence TREC.