Enterprise Resource Planning (ERP) systems are used to integrate all functions of an enterprise. They often evolve from a smaller monolithic object-oriented application, covering one functional area and organically grow over time in features and size until all functional areas are covered. Once they reach certain size, unrestricted dependencies among numerous classes increase complexity of the system and make it harder for development team to create new features and maintain code stability. This creates problems to further evolution of the ERP system and poses a risk to economic consequences for company developing it. ERP refactoring strategy, together with process of its creation, is presented in this thesis. It can be used with ERP systems, having architectural issues, with a purpose to improve quality of system’s architecture and thus prolong its lifecycle. The goal of modularizing monolithic system it pursued with intention to reduce complexity and make it easier to reason about the system. This architecture-level refactoring strategy is created for one specific medium-sized ERP system through iterative trial-and-error explorative approach. This thesis is carried out at the main development site for this ERP system by project team consisting of employees working on its development. The result shows the RefStratERP, an innovative refactoring strategy consisting of target architecture, refactoring process to reach it, refactoring principles and refactoring limitations. Contrary to initial expectation, arranging domain modules (modules containing business logic) in directed acyclic graph (DAG) is, in general, not feasible without sacrificing internal module cohesion of business logic. Accidental unidirectional dependency between two domain modules is at risk of becoming bidirectional under changing business requirements. On the other hand, non-domain modules (modules without business logic) could be completely separated from domain modules in a way that domain modules depend on non-domain modules. This comes from underlying nature of business domain and the fact that functional areas of an enterprise are interdependent. / Enterprise Resource Planning-system (ERP) används för att integrera alla funktioner inom ett företag. Oftast utvecklas dem från en mindre monolitisk objektorienterad applikation som täcker ett funktionellt område och växer organiskt över tiden i funktioner och storlek tills alla funktionella områden är täckta. När dem når en viss storlek ökar komplexiteten i systemet vilket gör det svårare fär utvecklingsteamet att skapa nya funktioner och hålla kodstabilitet. Detta skapar problem för fortsatt utveckling av ERP-systemet och utgör en risk för ekonomiska konsekvenser för utvecklingsföretaget. ERP refactoringstrategin, tillsammans med processen med att skapa den, presenteras i denna avhandling. Den kan användas med ERP-system, med arkitektoniska problem, med syfte att förbättra kvaliteten hos systemets arkitektur och därigenom förlänga dess livscykel. Målet att modularisera monolitiska system strävas efter i syfte att minska komplexiteten och göra det lättare att resonera kring systemet. En refaktorstrategi på arkitektnivå skapas för ett specifikt medelstort ERP-system genom en iterativ och försök-och-mistag-explorativ metod. Projektet genomfördes på ERPs huvudutvecklingsplats av ett projektteam bestående av anställda inom ERP utveckling. Resultatet visar RefStratERP, en innovativ refaktorstrategi som består av målarkitektur, refactoringprocess för att nå det, refactoringprinciper och refactoringbegränsningar. I motsats till inledande förväntningar är det i allmänhet inte möjligt att ordna domänmoduler (moduler som innehåller affärslogik) i en riktad acyklisk graf (DAG) utan att påverka interna modulsammanhang (cohesion) i affärslogiken. Oavsiktlig enriktad beroende mellan två domänmoduler riskerar att bli dubbelriktad under förändrade affärsbehov. Å andra sidan kan icke-domänmoduler (moduler utan affärslogik) helt separeras från domänmoduler så att domänmoduler beror på icke-domänmoduler. Detta kommer från underliggande egenskaper av affärsområden och det faktum att verksamhetsområden inom ett företag är beroende av varandra.

The subject of the dissertation is exploring the use of enterprise architecture as a managerial method to support the transformation (reform) of Czech public administration. The dissertation shows that enterprise architecture is in the public sector of Czech Republic still not used, even though public sector organizations want to continue improving its services and management in a way where enterprise architecture could be beneficial. The thesis aims to understand the needs of public authorities towards the possibilities and benefits of enterprise architecture and propose such a customization that could increase the adoption and use of enterprise architecture by these organizations. The main objective of thesis is to summarize individual partial results of author's research into the design of overall Concept of the structure and procedure of implementation of National Public Administration Architecture of Czech Republic. Research within the dissertation combines two basic research methods. Both represent qualitative research methods, suitable for the development of new methods and artefacts (Design Science Research) and for proof the hypotheses and outputs (Case Studies). To create preconditions of the proposed concept was necessary to verify the status of the Czech public administration from the perspective of current rate, needs and interest for the use of enterprise architecture to support the reform of public administration. To verify the status of environment and the needs served basic research questions, further elaborated into detailed questions of multiple case study. In the analytical part of the thesis are summarized significant findings the analysis of information resources and literature in the areas of a) the definition of enterprise architecture, its position, role and development trends for design changes in the structure and metamodel of TOGAF framework, b) the existence of available models and key design principles of application architecture reference model and c) experience of the implementation and use of enterprise architecture for public administration reform. The main part of the thesis focuses on the proposals in three areas. In the first part are designed theoretical and practical changes in the definition, structure and metamodel of enterprise architecture to better support its application for real reform of public administration, not only to improve the management of IT. Presented proposals are in the same time providing solutions to some identified discrepancies in the trends of development of enterprise architecture. General proposals are in parallel applied as changes in the standard TOGAF, which is then recommended as the initial framework for the National Public Administration Architecture of the Czech Republic. In the second part of the proposals are presented principles and content of the reference model of application architecture. This is an example the accelerator necessary to solve the discrepancy between the scope of enterprise architecture and its feasibility with limited resources. The thesis emphasize the industry adjusted model for public administration and its application in a hierarchical, fractal environments typical of public administration. In the third part are presented key components of the overall concept, using as well as previous research results of this thesis. As the key elements are presented proposal for the structure of National Public Administration Architecture of the Czech Republic environment, proposal for the procedure of implementation of National Architecture and the proposal of products (outputs) for deploying National Architecture in Czech Republic. Additionally are formulated proposals of ways of using National Architecture in Czech public administration and conditions and prerequisites for introduction of National Architecture in the Czech public administration.

Récemment, l'intégration entre les environnements informatiques et de réseautage a été largement promu pour fournir des services intelligents à des utilisateurs finaux ainsi que l'utilisation efficace des ressources. Cette convergence a ouvert la voie à l'émergence de l'internet des objets (IdO). Le paradigme de l'IdO repose principalement sur la fabrication d'objets, appelés les choses, disparaître et se tissent dans le tissu de notre vie de tous les jours pour nous soutenir dans l'accomplissement des activités quotidiennes. L'évolutivité et l'hétérogénéité sont parmi les principaux défis qui entravent la réalisation à grande échelle de services de l'IdO dans la vie quotidienne des utilisateurs. Afin de relever les défis de l'IdO, un nouveau volet de recherche est venu en avant dans la littérature comme une classe paradigmatique des Cyber-physiques systèmes sociaux (CSPR), qui est connu comme l'Internet social des choses (Siot). Le SIOT se fonde sur la notion soulignée par phénomène petite-monde où la structure sociale permettant relation sociale fondée sur la confiance entre les personnes et les objets, d'une manière qui ressemble à des services de réseaux sociaux traditionnels (SNS) est suggérée de relever les défis de l'IdO. Cependant, depuis SIOT hérite des caractéristiques de différents informatiques et de réseautage environnements (par exemple, l'IdO et SNS) cela, en fait, augmente la quantité et la variété des données contextuelles qui doit être manipulé pour Adaptive fourniture de services dans Siot, qui agit comme le principal défi adressé dans cette thèse. Autrement dit, dans cette thèse, nous proposons la notion de contexte cognitif lorsque, dans certaine situation spatio-temporelle, le raisonnement sur les aspects objectifs du cadre, ce qui représente l'environnement physique, avec le contexte subjective, qui représente les aspects comportementaux et sociaux, est considéré comme l'amélioration des services SIOT intelligence et la capacité d'adaptation aux besoins conjoncturels des utilisateurs. Nous envisageons technologies du Web sémantique pour déployer notre contexte cognitif proposé dans deux domaines d'application; sensible au contexte recommandation des tâches quotidiennes dans les maisons intelligentes et structure sociale dépendant de la situation des choses. Un prototype de preuve de concept a été développé pour chaque domaine d'application, dans le but de démontrer l'intégration harmonieuse des objets sur le Web pour la réalisation de certaines applications. Nos résultats empiriques montrent un niveau de service amélioré l'adaptabilité et la complexité en temps de fonctionner lors de l'application de notre contexte cognitif suggéré / The paradigm of the Social Internet of Things (SIoT) is being promoted in the literature to boost a new trend wherein the benefits of social network services are exhibited within the network of connected objects i.e., the Internet of Things (IoT). The novel user-friendly interaction framework of the SIoT opens the doors for enhancing the intelligence required to stimulate a shift in the IoT from a heterogeneous network of independently connected objects towards a manageable network of everything. In practice, achieving scalability within the large-scale and the heterogeneous paradigm of the IoT while maintaining on top of its user-friendly and intuitive services to bridge human-to-machine perceptions and encourage the technology’s adaptation is a major challenge which is hindering the realization and deployment of the IoT technologies and applications into people’s daily live. For the goal of handling IoT challenges, as well as improve the level of smart services adaptability to users’ situational needs, in this thesis, novel SIoT-based application architecture is provided. That is, Semantic Web Technologies are envisaged as a means to develop automated, value-added services for SIoT. While, interoperability and automation are essential requirement to seamlessly integrate such services into user life, Ontologies are used to semantically describe Web services with the aim of enabling the automatic invocation and composition of these services as well as support interactions across the cyber, physical and social worlds. On the other hand, handling the variety of contextual data in SIoT for intelligent decision making is another big challenge which is still in very early stages of research. In this thesis we propose a cognitive reasoning approach taking into consideration achieving situational-awareness (SA) in SIoT. This reasoning approach is deployed within two application domains where results show an improved level of services adaptability compared to location-aware services which are previously proposed in the literature