Os lapsos de fala são enunciados que apresentam um desvio em relação ao que o falante pretendia dizer (cf. FROMKIN, 1973). Esse erro pode se manifestar em níveis linguísticos distintos, afetando segmentos fonológicos, morfemas, palavras ou sentenças. Muitos linguistas defendem que os lapsos de fala são uma importante evidência para a postulação de modelos de performance que levem em conta a realidade psicológica das unidades linguísticas e suas regras (cf. FROMKIN, 1973; PFAU, 2009; LEVELT, 1989). O objetivo do nosso trabalho é, portanto, investigar a estrutura dos lapsos de fala espontâneos do português brasileiro que afetam o nível morfológico e explicitar as propriedades que regem os lapsos na nossa língua. Para isso, adotamos como modelo teórico o sistema de Pfau (2009), que foi desenvolvido com base nas evidências fornecidas pelos lapsos de fala e tem como alicerces a arquitetura da gramática proposta no quadro teórico da Morfologia Distribuída (cf. HALLE; MARANTZ, 1993; MARANTZ, 1997) e o modelo de performance em níveis (cf. LEVELT, 1989). Nossa hipótese inicial é a de que o sistema proposto por Pfau (2009) consegue dar conta da análise dos lapsos de fala em português brasileiro. Os resultados da pesquisa incluem a construção de um corpus constituído por 140 dados, coletados pelo método naturalístico, isto é, os lapsos foram anotados em contexto natural e espontâneo de fala, sem indução por testes. Os dados apontam para a existência de diferentes estruturas subjacentes aos lapsos de fala morfológicos no português do Brasil, sendo possível classificá-los em seis tipos distintos de acordo com suas características. As classes são: (1) blends de palavra, (2) blends frasais, (3) substituições que envolvem aspectos semânticos, (4) substituições que envolvem aspectos fonológicos, (5) lapsos que envolvem morfemas, e (6) lapsos que atingem traços gramaticais. / The slips of the tongue are utterances that present a deviation as compared to the speaker\'s intention (cf. FROMKIN, 1973). This error can occur in different linguistic levels, affecting phonological segments, morphemes, words or sentences. Many linguists claim that the slips of the tongue constitute important evidence for proposing performance models that consider the psychological reality of the linguistic units and the rules that apply to them. Our aim is to investigate the properties of spontaneous slips of the tongue in Brazilian Portuguese and to determine the structure of those which affect the morphological level. Hence, we have adopted as a theoretical model the system proposed in Pfau (2009), which has been developed based on evidence from the slips of the tongue. This system takes as a starting point the architecture of the grammar proposed by the Distributed Morphology model for language treatment (cf. HALLE; MARANTZ, 1993; MARANTZ, 1997) and by a model based on performance levels (cf. LEVELT, 1989). Our initial hypothesis is that Pfau\'s system can manage the analysis for slips of the tongue in Brazilian Portuguese. The results of this research include a corpus formed by 140 data, collected according to a naturalistic method, noted under natural context and spontaneous speech, without any induction by test. The data point to the existence of different subjacent structures to the morphological slips of the tongue in Brazilian Portuguese, classified in six distinct categories according to their characteristics. These types can be described as: (1) word blends, (2) phrasal blends (3) substitutions based on semantic aspects, (4) substitutions based on phonological aspects, (5) slips involving morphemes and, (6) slips involving grammatical features.

Os lapsos de fala são enunciados que apresentam um desvio em relação ao que o falante pretendia dizer (cf. FROMKIN, 1973). Esse erro pode se manifestar em níveis linguísticos distintos, afetando segmentos fonológicos, morfemas, palavras ou sentenças. Muitos linguistas defendem que os lapsos de fala são uma importante evidência para a postulação de modelos de performance que levem em conta a realidade psicológica das unidades linguísticas e suas regras (cf. FROMKIN, 1973; PFAU, 2009; LEVELT, 1989). O objetivo do nosso trabalho é, portanto, investigar a estrutura dos lapsos de fala espontâneos do português brasileiro que afetam o nível morfológico e explicitar as propriedades que regem os lapsos na nossa língua. Para isso, adotamos como modelo teórico o sistema de Pfau (2009), que foi desenvolvido com base nas evidências fornecidas pelos lapsos de fala e tem como alicerces a arquitetura da gramática proposta no quadro teórico da Morfologia Distribuída (cf. HALLE; MARANTZ, 1993; MARANTZ, 1997) e o modelo de performance em níveis (cf. LEVELT, 1989). Nossa hipótese inicial é a de que o sistema proposto por Pfau (2009) consegue dar conta da análise dos lapsos de fala em português brasileiro. Os resultados da pesquisa incluem a construção de um corpus constituído por 140 dados, coletados pelo método naturalístico, isto é, os lapsos foram anotados em contexto natural e espontâneo de fala, sem indução por testes. Os dados apontam para a existência de diferentes estruturas subjacentes aos lapsos de fala morfológicos no português do Brasil, sendo possível classificá-los em seis tipos distintos de acordo com suas características. As classes são: (1) blends de palavra, (2) blends frasais, (3) substituições que envolvem aspectos semânticos, (4) substituições que envolvem aspectos fonológicos, (5) lapsos que envolvem morfemas, e (6) lapsos que atingem traços gramaticais. / The slips of the tongue are utterances that present a deviation as compared to the speaker\'s intention (cf. FROMKIN, 1973). This error can occur in different linguistic levels, affecting phonological segments, morphemes, words or sentences. Many linguists claim that the slips of the tongue constitute important evidence for proposing performance models that consider the psychological reality of the linguistic units and the rules that apply to them. Our aim is to investigate the properties of spontaneous slips of the tongue in Brazilian Portuguese and to determine the structure of those which affect the morphological level. Hence, we have adopted as a theoretical model the system proposed in Pfau (2009), which has been developed based on evidence from the slips of the tongue. This system takes as a starting point the architecture of the grammar proposed by the Distributed Morphology model for language treatment (cf. HALLE; MARANTZ, 1993; MARANTZ, 1997) and by a model based on performance levels (cf. LEVELT, 1989). Our initial hypothesis is that Pfau\'s system can manage the analysis for slips of the tongue in Brazilian Portuguese. The results of this research include a corpus formed by 140 data, collected according to a naturalistic method, noted under natural context and spontaneous speech, without any induction by test. The data point to the existence of different subjacent structures to the morphological slips of the tongue in Brazilian Portuguese, classified in six distinct categories according to their characteristics. These types can be described as: (1) word blends, (2) phrasal blends (3) substitutions based on semantic aspects, (4) substitutions based on phonological aspects, (5) slips involving morphemes and, (6) slips involving grammatical features.

O recente crescimento da utilização de Unidades de Processamento Gráfico (GPUs) em aplicações científicas, que são voltadas ao desempenho, gerou a necessidade de otimizar os programas que nelas rodam. Uma ferramenta adequada para essa tarefa é o modelo de desempenho que, por sua vez, se beneficia da existência de uma ferramenta de extração de informações de desempenho para GPUs. Este trabalho cobre a criação de um gerador de microbenchmark para instruções PTX que também obtém informações sobre as características do hardware da GPU. Os resultados obtidos com o microbenchmark foram validados através de um modelo simplificado que obteve erros entre 6,11% e 16,32% em cinco kernels de teste. Também foram levantados os fatores de imprecisão nos resultados do microbenchmark. Utilizamos a ferramenta para analisar o perfil de desempenho das instruções e identificar grupos de comportamentos semelhantes. Também testamos a dependência do desempenho do pipeline da GPU em função da sequência de instruções executada e verificamos a otimização do compilador para esse caso. Ao fim deste trabalho concluímos que a utilização de microbenchmarks com instruções PTX é factível e se mostrou eficaz para a construção de modelos e análise detalhada do comportamento das instruções. / The recent growth in the use of tailored for performance Graphics Processing Units (GPUs) in scientific applications, generated the need to optimize GPU targeted programs. Performance models are the suitable tools for this task and they benefits from existing GPUs performance information extraction tools. This work covers the creation of a microbenchmark generator using PTX instructions and it also retrieves information about the GPU hardware characteristics. The microbenchmark results were validated using a simplified model with errors rates between 6.11% and 16.32% under five diferent GPU kernels. We also explain the imprecision factors present in the microbenchmark results. This tool was used to analyze the instructions performance profile, identifying groups with similar behavior. We also evaluated the corelation of the GPU pipeline performance and instructions execution sequence. Compiler optimization capabilities for this case were also verified. We concluded that the use of microbenchmarks with PTX instructions is a feasible approach and an efective way to build performance models and to generate detailed analysis of the instructions\' behavior.

Made available in DSpace on 2015-04-22T22:10:56Z (GMT). No. of bitstreams: 1 mario sobral.pdf: 1031304 bytes, checksum: 17b78abcce30773ebc357d63a1e2c07b (MD5) Previous issue date: 2013-08-06 / The Circle of Quality Control consists on a group of employees who volunteer to support a company, whose goal is to identify, analyze and propose improvements to problems detected in work, besides focusing on professional development and improving the work environment. The company studied has increased focus on financial return of the QCC, having no tools to evaluate the attendant performances and the impact on the work environment In this context, this thesis developed a conceptual model to analyze the performance of the QCC program, at an electronic company located in the Industrial Pole of Manaus, by the perspective of the attendant .Therefore, a study was done based in the program used by the company, literature review and existing models. Based by the models and documentary research, an instrument for data collection was developed and applied together at 60 attendents (leaders and secretaries) in September 2012. From the final results, we developed a conceptual model and the main result has achieved good performance of the current program (80.6%), otherwise identifying further development dimensions related to the company and the work environment when compared with the development of the attendents. Were also identified several opportunities for improvement QCC program at the company. / O Círculo de Controle de Qualidade constitui-se de um grupo de colaboradores voluntários que com o suporte da empresa, têm por objetivo identificar, analisar e propor melhorias para problemas detectados no trabalho, além de focar no desenvolvimento profissional dos circulistas e na melhoria do ambiente de trabalho. A empresa em estudo tem maior foco no retorno financeiro do CCQ, não possuindo ferramenta para avaliar o seu desempenho em relação ao desenvolvimento do participante e do impacto na melhoria do ambiente de trabalho. Neste contexto, a presente dissertação desenvolveu uma proposta de modelo conceitual para analisar o desempenho do programa CCQ em uma empresa de eletroeletrônico no Polo Industrial de Manaus, sob a ótica do participante. Para tanto, foi realizado um estudo para levantar as características do programa CCQ utilizado pela empresa, além de revisão bibliográfica e de modelos existentes. Com base nos modelos e na pesquisa documental, um instrumento de coleta de dados foi desenvolvido e aplicado junto a 60 circulistas (líderes e secretários) em setembro de 2012. A partir da análise dos resultados, desenvolvemos o modelo conceitual e o principal resultado alcançado foi o bom desempenho geral do atual programa (80,6%), porém identificando maior desenvolvimento nas dimensões relacionadas à empresa e ao ambiente de trabalho quando comparados com o desenvolvimento dos circulistas. Também foram identificadas junto aos participantes diversas oportunidades de melhorias para o programa de CCQ da empresa.

Management of prawn trawling fisheries is a difficult task due to the competing interests of strongly motivated stakeholders and interest groups. This occurs because prawn trawling operations are technically complex, require large capital investments and exhibit high running costs while owners have limited property rights over the resources that they harvest. Prawn stocks are public resources and are managed with a view to provide maximum benefit to the broad community. Additionally their exploitation also involves the incidental capture of significant numbers of other animals of no commercial value (bycatch) and causes impacts on seabed morphologies, which are involved in many diverse ecosystem processes. At the policy level an intention to manage trawl fisheries in a comprehensive way is backed by a mandated approach that is designed to capture all of the above issues and interests. That approach is termed Ecological Sustainable Development (ESD). The work in this thesis is designed to produce a prediction tool for prawn trawling performance that is based on modelling the physical nature of prawn trawling activities. It is proposed that the resulting tool is essential for working to manage the multi-dimensional aspects of prawn trawling fisheries. Three discrete objectives for the thesis are; to expand and improve an existing Prawn Trawling Performance Model (PTPM) so that it is more accurate and relevant to a broader range of questions, to evaluate the capacity of the PTPM to predict the performance characteristics of real prawn trawling operations in terms of both engineering and catching performance and to investigate the problem space surrounding prawn trawl fisheries to identify and develop applications for the model. A rudimentary PTPM (Sterling 2000b) is expanded through the analysis of further empirical data collected for model and full-scale trawl gear. / ght area of improvement to the PTPM were considered and in all cases significant changes were made. The accuracy of the new form of the model is here tested by comparing performance predictions with measurements of trawling performance for a variety of industrial trawl systems operated in the Queensland East Coast Trawl Fishery and also through comparing predicted trawling performance with prawn catches returned for trawlers operating in the Northern Prawn Fishery over the years 1970 to 2000. In the first case, errors in predicting swept area rate, considered an important performance parameter, were less than 5%. Fine scale issues were explored using the available sea mal data and a number of areas of concern within the model are highlighted. These relate to accurately quantifying the forces involved in the interaction of the trawl gear with the seabed and accurately accounting for the interaction between components within trawl systems. In the second case, the results suggest that between 50% and 60% of the variation in the seasonal catching performance of trawlers in the NPF is explained by predictions of swept area rate derived by the PTPM from the available data for that fishery. A comprehensive survey of applications for the PTPM is conducted in context with approaching the management of prawn trawling fisheries using the principles of ESD as defined by the National Strategy for ESD (1992). The Northern Prawn Fishery is used as a case study to explore in finer detail applications for the PTPM. Issues arising from the implementation of some of the applications are discussed.

The performance of different pavement maintenance treatments were evaluated by investigating practical projects collected from Tennessee Pavement Management System (PMS) and Long Term Pavement Performance (LTPP) database. The influence of factors on the effectiveness, cost-effectiveness and cracking initiation of different treatment were evaluated by “Optime”, multiple linear regression and parametric survival analysis. Pavement roughness, pavement serviceability index (PSI) and the initiation time of cracking were used as pavement performance indicators. Investigation on the pavement maintenance projects in Tennessee by Optime and multiple linear regression analysis indicated that HMA overlay had the highest effectiveness, followed by mill & fill and micro surfacing. Due to the relatively low cost, micro surfacing was the most cost-effective treatment, followed by HMA overlay and mill & fill. The effectiveness and cost-effectiveness decreased with the increase of traffic level and pre-treatment pavement condition. Investigation on the LTPP resurfacing treatments indicated that thick overlay and milling reduced the roughness after rehabilitation. Thin overlay, high traffic level and poor pre-rehabilitation pavement condition increased the deterioration rate of new overlay. Using reclaimed asphalt material did not influence the treatment performance but was cost-effective in reducing the roughness of new overlay. For a certain deterioration rate, there was an optimized pre-rehabilitation roughness value or time for applying maintenance treatment. Survival analysis on the crack initiation of asphalt overlay indicated that high traffic level accelerated the initiation of cracking. Thick overlay delayed the initiation of cracking except for the non-wheel path longitudinal crack. Mill retarded the occurrence of the non-fatigue cracks, whereas severe freeze thaw condition accelerated the occurrence of the two types of cracking. Using 30% RAP accelerated the initiation of longitudinal fatigue crack on wheel path but did not cause serious fatigue problem. The performance curves of HMA resurfacing treatments used in Tennessee were calibrated by investigating the influence of different factors on the slopes and intercepts of post-treatment performance curves. The analysis indicated that pavement with high pre-treatment PSI, thick overlay and deep milling had low deterioration rate, whereas pavement with higher traffic level deteriorated faster.

The current research sets out to demonstrate a comprehensive approach to validate complex human performance models as applied to time-sensitive tasks. This document is divided into 4 sections. Section 1 (Chapters 1 – 3) outlines previous efforts in the literature that have attempted to validate complex human performance models in the field with an emphasis on manual control models, task network models, cognitive models and integrated architectures. Section 2 (Chapters 4 – 7) elaborates on a validation approach and applies it to a baseline model of a complex task in the air traffic control domain. Section 3 (Chapters 7-12) outlines the importance of adopting an iterative model development-model validation process and reports on the three model iterations in an attempt to improve the validity of the baseline model. Each model augmentation was validated using the same validation approach and measures that were defined in Section 2. Section 4 (Chapters 13-14) provides a discussion and interpretation of the model results and highlights contributions to the field of both model validation and the field of human performance modelling of complex systems.

A safe and efficient highway infrastructure is a critical component and a valuable asset in terms of its monetary value, as well as supporting the way of life and economic activities of the people it serves. In North America, performing maintenance, repair, and expansion of an aging highway infrastructure to a target level of performance while dealing with ever-increasing traffic demands creates a significant challenge in terms of road user safety and mobility. Much of the current highway infrastructure was built several decades ago and it is therefore requiring increasing levels of maintenance and rehabilitation. <p>The cost of delays resulting from traffic congestion induced by work zones is estimated to be more than $6 billion per year. Work zone related traffic fatalities exceed more than 1000 lost lives per year in North America. Work zone related fatalities account for approximately 2.8 percent of highway fatalities in United States and 1.3 percent in Canada. While overall fatal crash rates have been steadily decreasing in both Canada and United States, work zone related fatalities have not been decreasing. <p>Smart Work Zones are an emerging technology designed to improve the safety and mobility within work zones on highways. Smart Work Zones employ various technologies to monitor current traffic conditions and provide relevant information to road managers and road users on current traffic flow conditions and automatically provide guidance to motorists for safer and more efficient navigation of the work zone. <p>This research examined the effects of a Smart Work Zone deployment by modeling traffic flow with and without a Smart Work Zone at the case study site in North Carolina to provide inputs into a performance analysis framework. The quantification of benefits and costs related to the deployment of a Smart Work Zone was developed in a probabilistic analysis framework model. The performance was quantified in economic terms of expected benefit cost ratio and net value realized from the deployment of a Smart Work Zone. The model considers the cost of deployment and potential savings in terms of motorist safety (fatal and injury crash reduction) as well as improvements in traveler mobility including reductions in user delays, vehicle operating costs, and emissions.<p>The model output is a risk profile that provides a range of expected values and associated probabilities of occurrence to quantify the expected benefits while also taking into consideration the uncertainty of the most sensitive input variables. The uncertainty of input variables determined to be the most sensitive were those associated with the amount of user delay and the valuation of user delay. The next most sensitive inputs are those associated with the cost of deploying and operating the Smart Work Zone system. <p>The model developed in this research concurs with the approach and analysis used in other models for the analysis of transportation projects. The model developed in this research provides a tool that can be used for decision making regarding the deployment of a Smart Work Zone and comparison with other transportation project alternatives. The model employs a user definable approach that enables it to be adapted to the specific conditions of a diverse range of field state conditions and has the ability to interface with several traffic flow models. <p>When applied to a case study project on Interstate 95 in North Carolina, the model was found to be capable of providing useful and relevant results that correlated to observed performance. The case study represented one of many operating scenarios on the project, and is not necessarily representative of all the field state conditions occurring over the period of the entire deployment. <p>The model results included a sensitivity analysis that identified the sensitivity of the outcome to uncertainty in the input values and a risk analysis that quantified the uncertainty of the predictions. The findings indicated that, at a 95 percent confidence level, the expected benefit / cost ratio of deploying a Smart Work Zone system was between 1.2 and 11.9 and the net value was between $10,000 and $225,000 per month of operation. Approximately 94 percent of the expected benefits were from savings in user delay and the remainder from savings due to improved safety, reduced emissions, and reduced vehicle operating costs. The results indicate that when applied under appropriate conditions, Smart Work Zones have the potential to provide significant benefits to road users. Under heavily congested conditions, the diversion of even a small amount of traffic to a more efficient route can provide sizable travel time improvements for all traffic.<p>In summary, the model developed in this research was specifically developed to apply to Smart Work Zones, but in its general form could also be applied to other work zone traffic management applications. In the case study the model was applied to a single rural work zone, but the framework could be extended for an integrated analysis of multiple work zones and network analysis in an urban setting. The research provides a fundamental framework and model for the analysis of Smart Work Zones and a method to determine the sensitivity of the uncertainty of input values. The research also identifies areas for continued examination of the effects of Smart Work Zone deployment and the prediction of expected benefits.

Due to the imperative need to reduce the management costs, operators multiplex several concurrent applications in large datacenters. However, uncontrolled resource sharing between co-hosted applications often results in performance degradation problems, thus creating violations of service level agreements (SLAs) for service providers. Therefore, in order to meet per-application SLAs, per-application performance modeling for dynamic resource allocation in shared resource environments has recently become promising. We introduce Chorus, an interactive performance modeling framework for building application performance models incrementally and on the fly. It can be used to support complex, multi-tier resource allocation, and/or what-if performance inquiry in modern datacenters, such as Clouds. Chorus consists of (i) a declarative high-level language for providing semantic model guidelines, such as model templates, model functions, or sampling guidelines, from a sysadmin or a performance analyst, as model approximations to be learned or refined experimentally, (ii) a runtime engine for iteratively collecting experimental performance samples, validating and refining performance models. Chorus efficiently builds accurate models online, reuses and adjusts archival models over time, and combines them into an ensemble of models. We perform an experimental evaluation on a multi-tier server platform, using several industry- standard benchmarks. Our results show that Chorus is a flexible modeling framework and knowledge base for validating, extending and reusing existing models while adapting to new situations.

Due to the imperative need to reduce the management costs, operators multiplex several concurrent applications in large datacenters. However, uncontrolled resource sharing between co-hosted applications often results in performance degradation problems, thus creating violations of service level agreements (SLAs) for service providers. Therefore, in order to meet per-application SLAs, per-application performance modeling for dynamic resource allocation in shared resource environments has recently become promising. We introduce Chorus, an interactive performance modeling framework for building application performance models incrementally and on the fly. It can be used to support complex, multi-tier resource allocation, and/or what-if performance inquiry in modern datacenters, such as Clouds. Chorus consists of (i) a declarative high-level language for providing semantic model guidelines, such as model templates, model functions, or sampling guidelines, from a sysadmin or a performance analyst, as model approximations to be learned or refined experimentally, (ii) a runtime engine for iteratively collecting experimental performance samples, validating and refining performance models. Chorus efficiently builds accurate models online, reuses and adjusts archival models over time, and combines them into an ensemble of models. We perform an experimental evaluation on a multi-tier server platform, using several industry- standard benchmarks. Our results show that Chorus is a flexible modeling framework and knowledge base for validating, extending and reusing existing models while adapting to new situations.