Ever since the early 90¡¦s the global software market and knowledge accumulation have been growing at a rapid speed. Knowledge has become a critical asset for professional software service industry and therefore, the demands for knowledge management in this industry have been increased dramatically. This study presents a novel approach for knowledge management in structured systems analysis and design area for a professional software service firm. A case study and a prototype system were used to illustrate the feasibility and usability of the proposed method. These results provide a practical base and better understanding for the professional software firms when they implement a knowledge management system.

Les systèmes polynomiaux à plusieurs variables apparaissent naturellement dans de nombreux domaines scientifiques. Ces systèmes issus d'applications possèdent une structure algébrique spécifique. Une méthode classique pour résoudre des systèmes polynomiaux repose sur le calcul d'une base de Gröbner de l'idéal associé au système. Cette thèse présente de nouveaux outils pour la résolution de tels systèmes structurés, lorsque la structure est induite par l'action d'un groupe ou une structure monomiale particulière, qui englobent les systèmes multi-homogènes ou quasi-homogènes. D'une part, cette thèse propose de nouveaux algorithmes qui exploitent ces structures algébriques pour améliorer l'efficacité de la résolution de systèmes (systèmes invariant sous l'action d'un groupe ou à support dans un ensemble de monômes particuliers). Ces techniques permettent notamment de résoudre un problème issu de la physique pour des instances hors de portée jusqu'à présent. D'autre part, ces outils permettent d'améliorer les bornes de complexité de résolution de plusieurs familles de systèmes polynomiaux structurés (systèmes globalement invariant sous l'action d'un groupe abélien, individuellement invariant sous l'action d'un groupe quelconque, ou ayant leur support dans un même polytope). Ceci permet en particulier d'étendre des résultats connus sur les systèmes bilinéaires aux systèmes mutli-homogènes généraux. / Multivariate polynomial systems arise naturally in many scientific fields. These systems coming from applications often carry a specific algebraic structure.A classical method for solving polynomial systems isbased on the computation of a Gr\"obner basis of the ideal associatedto the system.This thesis presents new tools for solving suchstructured systems, where the structure is induced by the action of a particular group or a monomial structure, which include multihomogeneous or quasihomogeneous systems.On the one hand, this thesis proposes new algorithmsusing these algebraic structures to improve the efficiency of solving suchsystems (invariant under the action of a group or having a support in a particular set of monomials). These techniques allow to solve a problem arising in physics for instances out of reach until now.On the other hand, these tools improve the complexity bounds for solving several families of structured polynomial systems (systems globally invariant under the action of an abelian group or with their support in the same polytope). This allows in particular to extend known results on bilinear systems to general mutlihomogeneous systems.

Cloud computing and big data technology continue to revolutionize how computing and data analysis are delivered today and in the future. To store and process the fast-changing big data, various scalable systems (e.g. key-value stores and MapReduce) have recently emerged in industry. However, there is a huge gap between what these open-source software systems can offer and what the real-world applications demand. First, scalable key-value stores are designed for simple data access methods, which limit their use in advanced database applications. Second, existing systems in the cloud need automatic performance optimization for better resource management with minimized operational overhead. Third, the demand continues to grow for privacy-preserving search and information sharing between autonomous data providers, as exemplified by the Healthcare information networks. My Ph.D. research aims at bridging these gaps. First, I proposed HINDEX, for secondary index support on top of write-optimized key-value stores (e.g. HBase and Cassandra). To update the index structure efficiently in the face of an intensive write stream, HINDEX synchronously executes append-only operations and defers the so-called index-repair operations which are expensive. The core contribution of HINDEX is a scheduling framework for deferred and lightweight execution of index repairs. HINDEX has been implemented and is currently being transferred to an IBM big data product. Second, I proposed Auto-pipelining for automatic performance optimization of streaming applications on multi-core machines. The goal is to prevent the bottleneck scenario in which the streaming system is blocked by a single core while all other cores are idling, which wastes resources. To partition the streaming workload evenly to all the cores and to search for the best partitioning among many possibilities, I proposed a heuristic based search strategy that achieves locally optimal partitioning with lightweight search overhead. The key idea is to use a white-box approach to search for the theoretically best partitioning and then use a black-box approach to verify the effectiveness of such partitioning. The proposed technique, called Auto-pipelining, is implemented on IBM Stream S. Third, I proposed ǫ-PPI, a suite of privacy preserving index algorithms that allow data sharing among unknown parties and yet maintaining a desired level of data privacy. To differentiate privacy concerns of different persons, I proposed a personalized privacy definition and substantiated this new privacy requirement by the injection of false positives in the published ǫ-PPI data. To construct the ǫ-PPI securely and efficiently, I proposed to optimize the performance of multi-party computations which are otherwise expensive; the key idea is to use addition-homomorphic secret sharing mechanism which is inexpensive and to do the distributed computation in a scalable P2P overlay.

La résolution de systèmes polynomiaux est un problème aux multiples applications, et les bases de Gröbner sont un outil important dans ce cadre. Il est connu que de nombreux systèmes issus d'applications présentent une structure supplémentaire par rapport à des systèmes arbitraires, et que ces structures peuvent souvent être exploitées pour faciliter le calcul de bases de Gröbner.Dans cette thèse, on s'intéresse à deux exemples de telles structures, pour différentes applications. Tout d'abord, on étudie les systèmes homogènes avec poids, qui sont homogènes si on calcule le degré en affectant un poids à chaque variable. Cette structure apparaît naturellement dans de nombreuses applications, dont un problème de cryptographie (logarithme discret). On montre comment les algorithmes existants, efficaces pour les polynômes homogènes, peuvent être adaptés au cas avec poids, avec des bornes de complexité générique divisées par un facteur polynomial en le produit des poids.Par ailleurs, on étudie un problème de classification de racines réelles pour des variétés définies par des déterminants. Ce problème a une application directe en théorie du contrôle, pour l'optimisation de contraste de l'imagerie à résonance magnétique. Ce système particulier s'avère insoluble avec les stratégies générales pour la classification. On montre comment ces stratégies peuvent tirer profit de la structure déterminantielle du système, et on illustre ce procédé en apportant des réponses aux questions posées par le problème d'optimisation de contraste. / Polynomial system solving is a problem with numerous applications, and Gröbner bases are an important tool in this context. Previous studies have shown that systèmes arising in applications usually exhibit more structure than arbitrary systems, and that these structures can be used to make computing Gröbner bases easier.In this thesis, we consider two examples of such structures. First, we study weighted homogeneous systems, which are homogeneous if we give to each variable an arbitrary degree. This structure appears naturally in many applications, including a cryptographical problem (discrete logarithm). We show how existing algorithms, which are efficient for homogeneous systems, can be adapted to a weighted setting, and generically, we show that their complexity bounds can be divided by a factor polynomial in the product of the weights.Then we consider a real roots classification problem for varieties defined by determinants. This problem has a direct application in control theory, for contrast optimization in magnetic resonance imagery. This specific system appears to be out of reach of existing algorithms. We show how these algorithms can benefit from the determinantal structure of the system, and as an illustration, we answer the questions from the application to contrast optimization.