Engineering contemporary distributed software applications is a challenging task due to the dynamic operating conditions in which these systems have to function. Examples are dynamic availability of resources, errors that are difficult to predict, and changing user requirements. These dynamics can affect a number of quality concerns of a system, such as robustness, openness, and performance. The challenges of engineering software systems with such dynamics have motivated the need for self-adaptation. Self-adaptation is based on the principle of separation of concerns, distinguishing two well defined systems: a managed system that deals with domain specific concerns and a managing system that deals with particular quality concerns of the managed system through adaptation with a feedback loop. State of the art in self- adaptation advocates the use of formal methods to specify and verify the system's behavior in order to provide evidence that the system's goals are satisfied. However, little work has been done on the consolidation of design knowledge to model and verify self-adaptation behaviors. To support designers, this thesis contributes with a set of formally specified templates for the specification and verification of self-adaptive behaviors of a family of distributed self-adaptive systems. The templates are based on the MAPE-K reference model (Monitor-Analyze-Plan-Execute plus Knowledge). The templates comprise: (1) behavior specification patterns for modeling the different MAPE components of a feedback loop, and (2) property specification patterns that support verification of the correctness of the adaptation behaviors. The target domain are distributed applications in which self-adaptation is used for managing resources for robustness and openness requirements. The templates are derived from expertise with developing several self-adaptive systems, including a collaborative mobile learning application in which we have applied self-adaptation to make the system robust to degrading GPS accuracy, and a robotic system in which we apply self-adaptation to support different types of openness requirements. We demonstrate the reusability of the templates in a number of case studies. / AMULETS

Business processes can be divided into standardisable and non-standardisable processes. Such processes are characterised by their activities, events, states and time-points. The conditions in which process activities, events, states and time-points occur determines how they are observed, recorded and acted upon. Under predictable and stable conditions an observer can record them by predetermining them and this can be done using existing process design approaches and methodologies. Such processes are termed standardisable processes; for example a process for manufacturing cars. However, under unpredictable and unstable conditions an observer cannot record them by predetermining all the possible events, because uncertainty leads to unpredictable events occurring. Such processes are termed non-standardisable processes and in this thesis as emergent processes; for example special engineering projects like building the Channel Tunnel. Therefore, a new approach is required for designing non-standardisable processes. Process events are significant because the observer notices and records them. Significantly, the observer of events also has to act on them. Whether process events occur in predictable and stable conditions or unpredictable and unstable conditions, makes a difference to how an observer notices, records and acts on them. This is highly significant for this thesis argument, because as non-standardisable process events can be unexpected and unpredictable or emergent a new approach is required to design them. This thesis advances knowledge of designing non-standardisable processes by conceptualising them as emergent business processes (EBP) and contributing a new approach for designing them using action research and the deferred design approach as a process design methodology. Uncertain and unpredictable conditions is characterised here as emergence. The observer cannot predetermine all the possible process events for processes that operate in emergent conditions and cannot determine how to act upon unpredictable process events, because some events will be predictable and others unpredictable. The pragmatist research methodology was used to research to identify and resolve the problem with EBP in the Foreign Procurement Division (FPD) of the Kuwait Ministry of Defence. It was also used an approach for designing EBP. The research contributes the new understanding of non-standardisable processes as emergent business processes. This is a significant contribution because it is conceptualisation that is not found in the literature. This conceptualisation recognises the need to find new approaches for designing and implementing EBP. Therefore, the research also contributes a new approach for designing EBP using the action research methodology as a process design methodology.

Computação em nuvem refere-se a um modelo de disponibilização de recursos computacionais no qual a infraestrutura de software e hardware é ofertada como um serviço, e vem se estabelecendo como um paradigma de sucesso graças a versatilidade e ao custo-efetividade envolvidos nesse modelo de negócio, possibilitando o compartilhamento de um conjunto de recursos físicos entre diferentes usuários e aplicações. Com o advento da computação em nuvem e a possibilidade de elasticidade dos recursos computacionais virtualizados, a alocação dinâmica de recursos vem ganhando destaque, e com ela as questões referentes ao estabelecimento de contratos e de de qualidade de serviço. Historicamente, as pesquisas em QoS concentram-se na solução de problemas que envolvem duas entidades: usuários e servidores. Entretanto, em ambientes de nuvem, uma terceira entidade passa a fazer parte dessa interação, o consumidor de serviços em nuvem, que usa a infraestrutura para disponibilizar algum tipo de serviço aos usuários finais e que tem recebido pouca atenção das pesquisa até o momento, principalmente no que tange ao desenvolvimento de mecanismos automáticos para a alocação dinâmica de recursos sob variação de demanda. Este trabalho consiste na proposta de uma arquitetura de gerenciamento adaptativo de recursos sob a perspectiva do modelo de negócio envolvendo três entidades, focada na eficiência do consumidor. O trabalho inspira-se em técnicas de controle realimentado para encontrar soluções adaptativas aos problemas de alocação dinâmica de recursos, resultando em uma arquitetura de broker de consumidor, um respectivo protótipo e um método de projeto de controle para sistemas computacionais dessa natureza / CLoud computing refers to a computer resource deployment model in which software and hardware infrastructure are offered as a service. Cloud computing has become a successful paradigm due to the versatility and cost-effectiveness involved in that business model, making it possible to share a cluster of physical resources between several users and applications. With the advent of cloud computing and the computer elastic resource, dynamic allocation of virtualized resources is becoming more prominent, and along with it, the issues concerning the establishment of quality of service parameters. Historically, research on QoS has focused on solutions for problems involving two entities: users and servers. However, in cloud environments, a third party becomes part of this interaction, the cloud consumer, that uses the infrastructure to provide some kind of service to endusers, and which has received fewer attention, especially regarding the development of autonomic mechanisms for dynamic resource allocation under time-varying demand. This work aims at the development of an architecture for dynamic adaptive resource allocation involving three entities, focused on consumer revenue. The research outcome is a consumer broker architecture based on feedback control, a respective architecture prototype and a computer system feedback control methodology which may be applied in this class of problems

Sistemas de recomendação foram propostos no início da década de 1990 com o objetivo de auxiliar seus usuários a lidar com a sobrecarga cognitiva criada com o advento da internet e o aumento constante de documentos. De lá para cá tais sistemas passaram a assumir vários outros papéis, tais como “auxiliar usuários a explorar”, “melhorar a tomada de decisão”, ou até mesmo “entreter”. Para atingir tais novos objetivos, o sistema necessita olhar para características do usuário que auxiliem no entendimento da tarefa desempenhada pelo usuário e como a recomendação pode auxiliar tal tarefa. Nesse sentido, propõe-se nessa tese uma integração entre estratégias de recomendação e de adaptação para criar um novo processo de recomendação adaptativa. É mostrado que tal integração pode melhorar a acurácia da recomendação, e dar bons resultados na retenção de usuários, e na interação destes com os sistemas. Para validar a abordagem, é implementado um protótipo para recomendação de filmes a serem utilizados em sala de aula. São também coletadas estatísticas de 78 usuários que participaram do experimento de avaliação da abordagem. / Recommender systems were proposed in early 90’s with the goal to help users deal with cognitive overload brought by the internet and the constant increase of documents. From there to now such systems have assumed many other roles like “help users to explore”, “improve decision making”, or even “entertain”. To accomplish such new goals, the system needs to look to user characteristics that help in understand what the user task is and how to adapt the recommendation to support such task. In this direction, it is proposed in this thesis an integration between recommender and adaptive strategies into a new process of adaptive recommendation. It is shown that such integration can improve recommendation accuracy and give good results to user retention, and interaction with the systems. To validate the approach, it is implemented a prototype to recommend movies to be used in a classroom. It is also collected some statistics about the 78 users who have participated of the experiment for evaluation of the new approach.

Esta é uma pesquisa quali-quantitativa que tem por base teórico-epistemológica o Interacionismo Sociodiscursivo (ISD) e a Teoria da Complexidade, uma vez que ambas as teorias contribuem para a compreensão da ação de linguagem, enquanto Sistema Adaptativo Complexo (SAC), de sujeitos engajados em atividades de produção de textos. Dessa forma, na medida em que o desempenho de sujeitos de diferentes níveis de ensino tem sido apontado como problemático por órgãos de pesquisa em Educação, no que se refere às práticas de leitura e de produção textual, considerou-se relevante investigar a maneira como uma ferramenta digital de mineração de dados, neste caso, a ferramenta Sobek, pode evidenciar níveis de complexidade entre o conteúdo de um texto-referência e o texto-produzido. O objeto de estudo desta tese, portanto, é o recorte propiciado pela ferramenta Sobek, que ocorre por meio da geração de um grafo composto por palavras (signos) recorrentes nos textos, com vistas a sustentar a construção de uma perspectiva inter(trans)disciplinar, referente aos modos de extrair e interpretar o conteúdo desses a partir de níveis de complexidade. O corpus de pesquisa foi composto por 2 turmas de ensino superior, em nível de graduação, envolvendo a coleta de 34 textos. Esses textos foram submetidos à técnica de mineração de dados, com a utilização de uma nova versão da ferramenta, desenvolvida neste estudo, a qual permitiu importar e comparar, de forma automatizada, o conteúdo de dois textos, sendo possível gerar 3 tipos de grafos: de união, de intersecção e de diferença. O resultado da análise de dados indicou um padrão de estabilidade da ação de linguagem dos sujeitos, em relação ao conteúdo do texto-referência, sem apresentar significativa variabilidade de conceitos. Para além de uma nova versão da ferramenta, outra contribuição desta pesquisa foi a de sugerir que níveis de complexidade em caráter de estabilidade e variabilidade podem ser associados a zonas de desenvolvimento humano. / This is a quali-quantitative research whose theoretical and epistemological basis are the Interactionism Sociodiscursivo (ISD) and the Complexity Theory, once both theories contribute to the understanding of the language action, as Complex Adaptive System (CAS), of individuals engaged in text production activities. Thus, as the individuals‟ performance in different levels of education has been identified as problematic by Research Institutes of Education, related to reading and textual production practices, it was considered relevant to investigate the way a digital data mining tool, in this case, Sobek, may reveal levels of complexity between the content of a reference-text and a produced-text. The object of study of this thesis, so is the cut provided by Sobek tool, which occurs through the generation of a graph composed of words (signs) recurring in the texts, in order to support the construction of an inter(trans) disciplinary perspective, referring to ways to extract and interpret the content of the texts. The research corpus was composed of 2 classes of higher education at the undergraduate level, involving the collection of 34 texts. These texts were submitted to data mining technique, with the use of a new version of the tool, developed in this study, which allowed to import and compare, automatically, the contents of two texts automatically, been possible to generate 3 types of graphs: union, intersection and difference. The result of data analysis indicated a recurring pattern in the analyzed texts, which was to maintain the language action as stable as possible to the content of the reference-text, without presenting significant variability of concepts. In addition to the new version of the tool, another contribution of this research was to suggest that the levels of complexity in stability and variability may be associated with human development zones.

Reducing hospital readmissions is critical to the success and sustainability of both hospitals and the communities in which they reside. The purpose of this multiple case study was to explore organizational strategies hospital leaders use to reduce hospital readmissions. The study was limited to hospitals in Southwest Missouri with readmission rates below the state average. Complex adaptive systems was the conceptual framework for the study because of the complex nature and numerous stakeholders of the healthcare system. Data were collected from a purposive sample of 15 hospital leaders via semistructured interviews and an analysis of organizational artifacts. Member checking was used to increase reliability and validity of the results. Data analysis was conducted using Yin's 5 step process including qualitative analysis software to identify major and core themes. The major themes identified in the study included population health, hospital operations and patient interactions, leadership and mission, and barriers to reducing readmissions. The implications for positive social change include the potential to improve services hospital team members provide to patients, which may improve the overall health of the communities they serve. By promoting improved health outcomes for local communities, society benefits through reduction of costs to the federal government and an overall improvement in the health of communities.

As the U.S. economy changed from industry to information, small cities suffered a decline in quality of life and an increase in poverty. The existing research has focused on demographics and descriptive attributes of micropolitan cities, but not on efforts to overcome these challenges. The purpose of this study was to explore and understand how a micropolitan city used economic and community development policies to rebuild its economy and improve quality of life. Using Holland's conceptualization of complex adaptive systems, research questions focused on triggers for policy creation and its use to create social change by improving the local economy and reducing the effects of poverty. Data for this qualitative case study were collected through open-ended questions in semi structured interviews with policymakers (elected officials), policy implementers (city employees), and policy influencers (community leaders). Interviews were supplemented with document review and photographic observation. The data were analyzed using descriptive coding, categorical aggregation, and direct interpretation to identify overarching themes of acceptance, resilience, building on strengths, and the interwoven nature of policy. The findings indicate that economic and community development policies can lead to positive changes such as the rehabilitation of blighted areas, growth of new and existing businesses, and influence state policy, illustrating the attributes of complex adaptive systems. The positive social change implications of this study include recommendations to city administrators to develop economic and community development policy based on their unique circumstances, to build partnerships, promote community change, and build a positive mindset to benefit their city and citizens.

The research question for this study is “Can concepts from complex adaptive systems and conversation analysis be used to research consumer behaviour?" This is, primarily, a theoretical question. After a wide-ranging literature search no scholarly publications linking the qualitative aspects of complex adaptive systems theory to marketing or consumer research were located. In addition, there appear to be few papers on consumer research which use conversation analysis. A theory for the research methodology was developed. It was argued that the production of a research theory and methodology to test the relevance and appropriateness of two very different theories - complex adaptive systems and conversation analysis was the major undertaking of this thesis. The problem of combining an essentially scientific perspective (complex adaptive systems) with an essentially qualitative one (ethnomethodology and conversation analysis) was resolved as part of the research process. A bridging theory was developed through the common ground offered by the sociology of scientific knowledge on the one hand and social-constructionist theory on the other. This methodology was successful in supporting the choice of conversation analysis as the data-collection method and provided the rationale for observing five characteristics of a complex adaptive system. The methodology was tested empirically and, in keeping with exploratory work, iteratively. It is not intended that this type of research will have predictive value. The complex adaptive system studied was consumers in a small group. There were two research locations with six data-collection sessions in each. The first location collected data from organisational groups. The second collected data from groups of consumers convened in a meeting room. / Data were transcribed and analysed for all sessions according to the conventions of conversation analysis. In the meeting-room sessions, data were also collected by electronic-group-support-systems technology and subjected to a modified form of content analysis. The broad findings showed the following. The assumption that there was little evidence of interest in complex adaptive systems among consumer behaviour researchers was confirmed. Apart from one paper calling for the use of conversation analysis in consumer behaviour research, there appeared to have been no subsequent reports of its adoption. The potential for conversation analysis in consumer research has probably not been understood because it was seen as a data-collection method only within an ethnomethodological perspective. The discursive theoretical perspective, which gives a prime position to conversation analysis in the construction of factual accounts, was found to be an innovative way to study consumer behaviour. A discursive theoretical research perspective could have provided a more robust theoretical justification for the fieldwork carried out in this study than the theory of the methodology that was first developed for this study. Conversation analysis did meet the five criteria proposed for surfacing a complex adaptive system in a small group but in an unexpected way. It met these criteria through the research process. In other words, by setting up an appropriate research environment and using conversation analysis, it was shown that a complex adaptive system was in operation. / An outcome of employing complex adaptive systems theory and conversation analysis is a new way of seeing groups of consumers as a self-organised, nonlinear, interactive entity. Conversation analysis has proven to be a method of empirically observing this entity, whilst preserving the consumer groups' complex adaptiveness. There were three conclusions. The first is that the discursive paradigm appears to be an alternative paradigm for consumer behaviour research that is appropriate for certain applications. For example, marketing communications and word-of-mouth communication. The second conclusion is that when small-group talk-in-interaction is recorded and analysed using conversation analysis, the characteristics of a complex adaptive system theorised in this study seem evident to the researcher. The third is that complex adaptive systems appear to be capable of being researched in the field, but more work is needed on defining the characteristics to be researched.

A principal purpose of this thesis is to present an economic evaluation of ecosystems. The concept of ecosystem health is adopted to ascertain the status of ecosystems. Ecosystem health is considered in part an economic concept and defined as a function of utility through the ecosystem services that satisfy various needs, subject to preserving the integrity of the adaptive cycle. In order to quantify the utility supplied by ecosystem services the novel utility index Ecosystem Outcome Protection Year (ECOPY) is developed. By forming this index, an evaluation can be performed using cost utility analysis, which avoids monetizing these benefits. An attempt is made to ascertain an appropriate approach for ecosystem management. It is reasoned that expert intuition can determine some kind of macro-regularities in ecosystems despite their complex dynamics. Hence, these inferences could be used for ecosystem management. Adaptive co-management is introduced as a means to bring about the collaboration of experts as resource co-managers. The concept of informed intuition is developed to bring about a systematic approach to learning and evaluation where the mental models of experts are transcribed using fuzzy cognitive mapping. However, it is argued that ecosystems as complex adaptive systems are non-ergodic and full of surprises. Accordingly, abduction, the logic of creative conjecture is systematically developed, for the purposes of maintaining mental model flexibility. This systematic application of abduction with an informed intuition forms the proposed abductive process of research, which is grounded in Shacklean potential surprise, a non-probabilistic function. To demonstrate this novel research process, a post-classical economic evaluation of Te Waihora lake ecosystem is undertaken, which employs the ECOPY index and potential surprise method. This empirical case study reveals various cost-effective management actions for improving lake health, which went beyond the intuitions of resource co-managers. This indicated the potential of the approach, which is considered a significant contribution for the methodological development of ecosystem management.

The work presented in this dissertation investigates techniques for robot Learning from Demonstration (LFD). LFD is a well established approach where the robot is to learn from a set of demonstrations. The dissertation focuses on LFD where a human teacher demonstrates a behavior by controlling the robot via teleoperation. After demonstration, the robot should be able to reproduce the demonstrated behavior under varying conditions. In particular, the dissertation investigates techniques where previous behavioral knowledge is used as bias for generalization of demonstrations. The primary contribution of this work is the development and evaluation of a semi-reactive approach to LFD called Predictive Sequence Learning (PSL). PSL has many interesting properties applied as a learning algorithm for robots. Few assumptions are introduced and little task-specific configuration is needed. PSL can be seen as a variable-order Markov model that progressively builds up the ability to predict or simulate future sensory-motor events, given a history of past events. The knowledge base generated during learning can be used to control the robot, such that the demonstrated behavior is reproduced. The same knowledge base can also be used to recognize an on-going behavior by comparing predicted sensor states with actual observations. Behavior recognition is an important part of LFD, both as a way to communicate with the human user and as a technique that allows the robot to use previous knowledge as parts of new, more complex, controllers. In addition to the work on PSL, this dissertation provides a broad discussion on representation, recognition, and learning of robot behavior. LFD-related concepts such as demonstration, repetition, goal, and behavior are defined and analyzed, with focus on how bias is introduced by the use of behavior primitives. This analysis results in a formalism where LFD is described as transitions between information spaces. Assuming that the behavior recognition problem is partly solved, ways to deal with remaining ambiguities in the interpretation of a demonstration are proposed. The evaluation of PSL shows that the algorithm can efficiently learn and reproduce simple behaviors. The algorithm is able to generalize to previously unseen situations while maintaining the reactive properties of the system. As the complexity of the demonstrated behavior increases, knowledge of one part of the behavior sometimes interferes with knowledge of another parts. As a result, different situations with similar sensory-motor interactions are sometimes confused and the robot fails to reproduce the behavior. One way to handle these issues is to introduce a context layer that can support PSL by providing bias for predictions. Parts of the knowledge base that appear to fit the present context are highlighted, while other parts are inhibited. Which context should be active is continually re-evaluated using behavior recognition. This technique takes inspiration from several neurocomputational models that describe parts of the human brain as a hierarchical prediction system. With behavior recognition active, continually selecting the most suitable context for the present situation, the problem of knowledge interference is significantly reduced and the robot can successfully reproduce also more complex behaviors.