xii, 55 p. : ill. A print copy of this thesis is available through the UO Libraries. Search the library catalog for the location and call number. / Selective attention refers to perceptual selection and working memory refers to the active maintenance of mental representations. Selective attention and working memory are believed to be two of the most important functions in human cognition and have been intensively investigated in cognitive psychology. However, it is quite recent that the link between attention and working memory has been systematically researched. One question that remains controversial is the effect of working memory on attentional control with inconsistent results reported in the human psychophysical literature, despite clear and strong evidence from physiological studies with nonhuman primates that working memory is the main source of top-down attentional control. The main goal of the current study is to provide a plausible solution to the puzzle of attentional control by introducing the concept of goal-specificity and competition between working memory representations. I hypothesized that the strength of the biasing effect of working memory on attention depends on the specificity of representations in working memory, and developed an experimental paradigm (the goal-specificity paradigm) to test this hypothesis using psychophysical and neuroimaging methods. One of the most important manipulations in the goal-specificity paradigm is how specifically targets in different tasks are defined. The results demonstrate that there is competition between items in working memory for attentional control that is influenced by the specificity of each representation as well as task relevancy. Also, it is shown that the effect of goal-specificity is present in both spatial and temporal domains as revealed by visual search and rapid serial visual presentation tasks. The results suggest the possibility that the negligible effect of working memory in some previous studies may be due to insufficient specificity of the objects in working memory or to the presence of other specifically-defined information in working memory. Furthermore, based on the implication from the current study that goal-specificity has a significant influence on attentional control, I expect that the experimental paradigm introduced in the current study can be utilized as an objective psychophysical measure of attentional control. / Committee in charge: Margaret Sereno, Chairperson, Psychology; Scott Frey, Member, Psychology; Michael Wehr, Member, Psychology; Richard Taylor, Outside Member, Physics

Made available in DSpace on 2014-06-11T19:26:07Z (GMT). No. of bitstreams: 0 Previous issue date: 2004-02-18Bitstream added on 2014-06-13T20:14:40Z : No. of bitstreams: 1 pimentel_ecg_me_jabo_prot.pdf: 1045293 bytes, checksum: 7125db7c779353ee29e1494c9a78d178 (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / O problema da multicolinearidade em análises de regressão foi abordado. A técnica da regressão de cumeeira foi empregada na estimação de parâmetros de efeitos genéticos sobre o desempenho de animais cruzados, por meio de dois modelos: um contendo apenas efeitos de ação aditiva e dominância (AD); e outro incluindo, além desses, efeitos de epistasia e complementariedade (ADEC). Um programa foi desenvolvido, em linguagem Fortran 90, para implementação de cinco versões da regressão de cumeeira: o método proposto originalmente; o implementado pelo SAS; e três formas de ponderação do coeficiente ë. Três critérios matemáticos para escolha de ë foram testados: a soma e a média harmônica dos valores absolutos da estatística t de Student, e o valor de ë a partir do qual os valores dos fatores de inflação de variância passavam a ser todos menores que trezentos. As comparações entre os cinco métodos e os três critérios foram feitas, usando-se o modelo ADEC, pelo exame de superfícies de predição obtidas a partir dos coeficientes estimados. Superfícies de predição também foram usadas para comparação entre os dois modelos, para cada método. Com o conjunto de dados utilizado, superfícies de predição biologicamente coerentes puderam ser obtidas em todos os métodos de implementação, usando-se o critério com base nos valores de FIV para determinação de ë. Recomenda-se que um critério matemático seja usado como ferramenta auxiliar para escolha de ë, não dispensando o exame dos sinais e valores das estimativas e um bom conhecimento do fenômeno em estudo. A inclusão de parâmetros para efeitos de epistasia e complementariedade em modelos de avaliação de efeitos genéticos em animais cruzados pôde representar um ganho tanto em termos de ajuste do modelo quanto de capacidade de predição de desempenho de genótipos não testados. / The problem of multicollinearity in regression analysis was studied. Ridge regression techniques were used to estimate genetic parameters affecting performance of crossbred animals, using two models: the additive-dominance model; and an alternative model including additive, dominance, complementarity and epistatic effects. A software was developed, in Fortran 90, to perform five variant types of ridge regression: the originally proposed method; the one implemented by SAS; and three forms of weighting the ridge coefficient ë. Three mathematical criteria were tested with the aim of choosing a value for the ë coefficient: the sum and the harmonic mean of absolute Student t-values, and the value of ë from which all variance inflation factors (VIFs) became lower than 300. Prediction surfaces, obtained from estimated coefficients, were used to compare the five methods and three criteria, using the alternative model. Prediction surfaces were also used to compare the two models, for each method. In this study (and this particular data structure), prediction surfaces showed quite acceptable biological interpretation, for all five methods, when criterion based on VIF values was used to choose the ë coefficient. A mathematical criterion to choose ë is recommended as an indicator tool, without excluding an exam of signs and values of estimated coefficients, and a good understanding of the phenomenon under study. Inclusion of complementarity and epistatic effects, in models for genetic effects evaluation in crossbred animals, represented a better fit of the model, and an improvement in its ability to predict performance of untested genotypes.

A wireless sensor network (WSN) is made of tiny sensor nodes usually deployed in high density within a targeted area to monitor a phenomenon of interest such as temperature, vibration or humidity. The WSNs can be employed in various applications (e.g., Structural monitoring, agriculture, environment monitoring, machine health monitoring, military, and health). For each application area there are different technical issues and remedies. Various challenges need to be considered while setting up a WSN, including limited computing, memory and energy resources, wireless channel errors and network scalability. One way of addressing these problems is by implementing a routing protocol that efficiently uses these limited resources and hence reduces errors, improves scalability and increases the network lifetime. The topology of any network is important and wireless sensor networks (WSNs) are no exception. In order to effectively model an energy-efficient routing algorithm, the topology of the WSN must be factored in. However, little work has been done on routing for WSNs with regular patterned topologies, except for the shortest path first (SPF) routing algorithms. The issue with the SPF algorithm is that it requires global location information of the nodes from the sensor network, which proves to be a drain on the network resources. In this thesis a novel algorithm namely, BRALB (Biased Random Algorithm for Load Balancing) is proposed to overcome the issues faced in routing data within WSNs with regular topologies such as square-base topology and triangle-based topology. It is based on random walk and probability. The proposed algorithm uses probability theory to build a repository of information containing the estimate of energy resources in each node, in order to route packets based on the energy resources in each node and thus does not require any global information from the network. It is shown in this thesis by statistical analysis and simulations that BRALB uses the same energy as the shortest path first routing as long as the data packets are comparable in size to the inquiry packets used between neighbours. It is also shown to balance the load (i.e. the packets to be sent) efficiently among the nodes in the network. In most of the WSN applications the messages sent to the base station are very small in size. Therefore BRALB is viable and can be used in sensor networks employed in such applications. However, one of the constraints of BRALB is that it is not very scalable; this is a genuine concern as most WSNs deployment is large scale. In order to remedy this problem, C-BRALB (Clustered Biased Random Algorithm for Load Balancing) has been proposed as an extension of BRALB with clustering mechanism. The same clustering technique used in Improved Directed Diffusion (IDD) has been adopted for C-BRALB. The routing mechanism in C-BRALB is based on energy biased random walk. This algorithm also does not require any global information apart from the initial flooding initiated by the sink to create the clusters. It uses probability theory to acquire all the information it needs to route packets based on energy resources in each cluster head node. It is shown in this thesis by using both simulations and statistical analysis that C-BRALB is an efficient routing algorithm in applications where the message to be sent is comparable to the inquiry message among the neighbours. It is also shown to balance the load (i.e. the packets to be sent) among the neighbouring cluster head nodes.

Os receptores acoplados à proteína G (GPCRs), também chamados de receptores 7TM, são conhecidos por regular virtualmente todos os processos fisiológicos em mamíferos e cerca de 40% de todas as drogas comerciais agem através destes receptores. A sinalização mediada por eles é classicamente atribuída à proteína G, que é ativada pela troca de GDP por GTP, promovendo a separação das subunidades G e G, e leva à produção de mensageiros secundários como cAMP, Ca2+ e DAG. Após a resposta os GPCRs são fosforilados pelas quinases de GPCRs (GRKs), sinalizando para recrutamento das -arrestinas citoplasmáticas, que por sua vez desencadeiam a formação de endossomos internalizando e dessensibilizando o receptor. Entretanto, estudos mostram que este endossomo, contendo o complexo ligante-receptor--arrestina, pode interagir com proteínas sinalizadoras no citoplasma desencadeando vias de sinalização independentes de proteína G. Recentemente foram descritos para diferentes receptores, ligantes capazes de ativar seletivamente uma das duas vias, proteína G ou -arrestina, chamados agonistas seletivos. O receptor AT1 é um GPCR particularmente interessante no estudo do agonismo seletivo, tanto por sua vasta expressão em tecidos quanto pelo conhecimento de agonistas seletivos já estabelecidos, tais como os ligantes SII e TRV120027. O objetivo deste trabalho foi analisar comparativamente os perfis de sinalização decorrente da ativação de AT1 por SII ou TRV120027 através do uso de arranjos de quinases e da modulação de genes relacionados a sinalização de GPCRs. Ang II que é ligante natural e total (ativa via dependente de proteína G e de -arrestina) neste receptor foi usada como controle para fins de comparação. Nossos dados mostraram que o perfil da sinalização mediada pelo receptor AT1 varia não só entre AngII e os agonistas seletivos, mas também entre os dois ligantes seletivos SII e TRV120027, mostrando que a interação receptor-ligante pode influenciar a sinalização em um grau mais refinado, além da ativação dependente de -arrestina ou proteína G. Estes dados mostram que existem perspectivas para o desenvolvimento futuro de ligantes com ainda maior grau de seletividade. / G protein coupled receptors (GPCRs), also known as 7TM receptors, are known to regulate virtually all physiological processes in mammals and approximately 40% of all current clinical drugs act by modulating such receptors. The signaling mediated by them is classically by coupling to G protein, which is activated by exchanging bound GDP for GTP, dissociation of G and G subunits, then leading to production of second messengers such as cAMP, Ca2+, and DAG. After the signal transduction, GPCR are phosphorylated by GPCR kinases (GRKs), followed by recruitment of cytoplasmic -arrestins, which initiate the endosome formation with consequent internalization and desensitization of the receptor. However, is has been demonstrated that the endosome assembling the ligand-receptor--arrestin complex can interact with cytoplasmic signaling proteins, therefore activating signaling pathways independently of G protein coupling. Recently, for different receptors, it has been described ligands capable of selectively activating one of these signaling pathways, G protein or -arrestin, called biased agonists. The AT1 receptor is a particularly interesting GPCR for the study of biased agonism, either due to its wide tissue expression as well as also due the existence of known and established biased ligands, such as SII and TRV120027. The aim of our study was to comparatively analyze the AT1 receptor signaling pathways profiles after activation by SII or TRV120027, using kinases arrays, and expression modulation of genes related to GPCRs signaling. AngII is the natural and full agonist of this receptor (activates both G protein and -arrestin signaling pathways) was used for comparison. Our data show that the signaling profile mediated by AT1 receptor can be distinct not only when comparing the profiles from AngII and the biased agonists, but also when comparing the profiles from the two biased ligands SII and TRv120027; revealing that the complex ligand-receptor can influence the downstream signaling pathways in a fine-tune way, further to the activation of -arrestin or G-protein. This data show that there are perspectives for the future development of ligands with even higher degree of selectivity.

Resumo: O problema da multicolinearidade em análises de regressão foi abordado. A técnica da regressão de cumeeira foi empregada na estimação de parâmetros de efeitos genéticos sobre o desempenho de animais cruzados, por meio de dois modelos: um contendo apenas efeitos de ação aditiva e dominância (AD); e outro incluindo, além desses, efeitos de epistasia e complementariedade (ADEC). Um programa foi desenvolvido, em linguagem Fortran 90, para implementação de cinco versões da regressão de cumeeira: o método proposto originalmente; o implementado pelo SAS; e três formas de ponderação do coeficiente ë. Três critérios matemáticos para escolha de ë foram testados: a soma e a média harmônica dos valores absolutos da estatística t de Student, e o valor de ë a partir do qual os valores dos fatores de inflação de variância passavam a ser todos menores que trezentos. As comparações entre os cinco métodos e os três critérios foram feitas, usando-se o modelo ADEC, pelo exame de superfícies de predição obtidas a partir dos coeficientes estimados. Superfícies de predição também foram usadas para comparação entre os dois modelos, para cada método. Com o conjunto de dados utilizado, superfícies de predição biologicamente coerentes puderam ser obtidas em todos os métodos de implementação, usando-se o critério com base nos valores de FIV para determinação de ë. Recomenda-se que um critério matemático seja usado como ferramenta auxiliar para escolha de ë, não dispensando o exame dos sinais e valores das estimativas e um bom conhecimento do fenômeno em estudo. A inclusão de parâmetros para efeitos de epistasia e complementariedade em modelos de avaliação de efeitos genéticos em animais cruzados pôde representar um ganho tanto em termos de ajuste do modelo quanto de capacidade de predição de desempenho de genótipos não testados. / Abstract: The problem of multicollinearity in regression analysis was studied. Ridge regression techniques were used to estimate genetic parameters affecting performance of crossbred animals, using two models: the additive-dominance model; and an alternative model including additive, dominance, complementarity and epistatic effects. A software was developed, in Fortran 90, to perform five variant types of ridge regression: the originally proposed method; the one implemented by SAS; and three forms of weighting the ridge coefficient ë. Three mathematical criteria were tested with the aim of choosing a value for the ë coefficient: the sum and the harmonic mean of absolute Student t-values, and the value of ë from which all variance inflation factors (VIFs) became lower than 300. Prediction surfaces, obtained from estimated coefficients, were used to compare the five methods and three criteria, using the alternative model. Prediction surfaces were also used to compare the two models, for each method. In this study (and this particular data structure), prediction surfaces showed quite acceptable biological interpretation, for all five methods, when criterion based on VIF values was used to choose the ë coefficient. A mathematical criterion to choose ë is recommended as an indicator tool, without excluding an exam of signs and values of estimated coefficients, and a good understanding of the phenomenon under study. Inclusion of complementarity and epistatic effects, in models for genetic effects evaluation in crossbred animals, represented a better fit of the model, and an improvement in its ability to predict performance of untested genotypes. / Orientador: Luiz Alberto Fries / Coorientadora: Sandra Aidar de Queiroz / Banca: Maurício Mello de Alencar / Banca: Danísio Prado Munari / Mestre

The ratchet effect can be exploited in many types of research, yet few researchers pay attention to it. In this thesis, I investigate two examples of such effects in microfluidic devices, under the guidance of computational simulations. The first chapter provides a brief introduction to ratchet effects, electrophoresis, and swimming cells, topics directly related to the following chapters. The second chapter of this thesis studies the separation of charged spherical particles in various microfluidic devices. My work shows how to manipulate those particles with modified temporal asymmetric electric potentials. The rectification of randomly swimming bacteria in microfluidic devices has been extensively studied. However, there have been few attempts to optimize such rectification devices. Mapping such motion onto a lattice Monte Carlo model may suggest some new mathematical methods, which might be useful for optimizing the similar systems. Such a mapping process is introduced in chapter four.

L'angiotensine II (Ang II) joue un rôle important dans la régulation du système cardiovasculaire par l’activation de plusieurs voies de signalisation. L’activation de ces voies passe par le récepteur de l'angiotensine II de type 1 (AT1R). Ce récepteur fait partie de la famille des récepteurs couplés aux protéines G (GPCRs). De plus, il est maintenant connu que certains ligands peuvent lier le récepteur et induire une conformation qui permet d'activer certaines voies de signalisation tout en n’étant pas favorable à l'activation d'autres voies. Il est alors question de sélectivité fonctionnelle, aussi appelée signalisation biaisée. Ainsi, avec cette approche, il est possible de cibler les voies qui produiront les effets thérapeutiques désirés sans toutefois activer les voies qui seraient responsables des effets indésirables. Nous avons émis l’hypothèse que de cycliser des ligands va restreindre les conformations possibles lors du couplage avec AT1R et induire un agonisme biaisé. Ainsi, des analogues cycliques de l’AngII substitués aux positions 3 et 5 par des cystéines et des homocystéines ont été synthétisés: [Sar1Hcy3,5]AngII, [Sar1Cys3Hcy5]AngII et [Sar1Cys3,5]AngII. D’abord, la capacité de ces analogues cycliques à activer la voie Gq a été évaluée par la mesure de la production des inositol phosphates. Puis, la capacité à activer les voies G12, le recrutement des β-arrestines (1 et 2) ainsi que l’activation de ERK1/2 a également été évaluée. Nos travaux ont montré que l’analogue cyclique [Sar1Hcy3,5]AngII a une puissance et une efficacité maximales sur toutes les voies testées à l'exception de la voie Gq. Des simulations de dynamique moléculaire ont été effectuées pour nous permettre de comprendre comment la conformation du ligand influence la structure d’AT1R et donc l’activation des différentes voies de signalisation. Les simulations en dynamique moléculaire ont montré que la barrière énergétique associée à l'insertion du résidu Phe8 de l’AngII dans le coeur hydrophobe d'AT1R est augmentée avec [Sar1Hcy3,5]AngII, pouvant expliquer que cet analogue active moins bien la voie Gq. D’autres analogues cyclisés aux positions 3 et 5 de l’AngII ont été synthétisés; [Sar1Hcy3Ile4Hcy5]AngII, [Sar1Hcy3,5Ile8]AngII et [Sar1Hcy3Cys5]AngII. Leur capacité à activer les voies Gq, ERK1/2 et le recrutement des β-arrestines (1 et 2) a été évaluée. L’analogue [Sar1Hcy3Cys5]AngII semblait bien activer la voie ERK1/2, mais pas les voies G12 et β-arrestines. Ces résultats suggèrent que le fait de contraindre les mouvements des déterminants moléculaires d’un ligand en introduisant des structures cycliques peut entraîner un biais dans la signalisation en stabilisant différentes structures du récepteur. / Abstract: Angiotensin II (Ang II) has an important role in the regulation of the cardiovascular system by its ability to activate several signaling pathways. The activation of these pathways occurs via the angiotensin II receptor type 1 (AT1R). This receptor belongs to the family of G protein-coupled receptors (GPCRs). Moreover, it is now known that certain ligands can bind to the receptor and induce a conformation that allow the activation of certain signaling pathways while not promoting the activation of other pathways. This concept is known as functional selectivity or biased signaling. With this approach, it is possible to target the signaling pathways that produce the desired therapeutic effects rather than activating the pathways responsible for adverse effects. We hypothesized that cyclizing ligands would restrict possible conformations when coupled with AT1R and induce biased agonism. Thus, cyclic AngII analogs substituted at positions 3 and 5 by cysteines and homocysteines were synthesized: [Sar1Hcy3,5]AngII, [Sar1Cys3Hcy5]AngII and [Sar1Cys3,5]AngII. First, the ability of these cyclic analogs to activate the Gq pathway was measured by the inositol phosphates production. Then, the G12 pathway activation, β-arrestin (1 and 2) recruitment and the ability of these analogs to activate the ERK1/2 pathway was evaluated. Our work has shown that [Sar1Hcy3,5]AngII has maximum potency and efficacy on all of the evaluated pathways, except for the Gq pathway. Molecular dynamic simulations were used to understand how a distinct ligand conformation influences the AT1R structure and the activation of signaling pathways. These studies have shown that the energy barrier associated with the insertion of the Phe8residue of AngII within the hydrophobic core of AT1R is increased with [Sar1Hcy3,5]AngII, possibly explaining why this analog is less potent in activating the Gq pathway. Other analogues cyclized at positions 3 and 5 of AngII were synthesized; [Sar1Hcy3Ile4Hcy5]AngII, [Sar1Hcy3,5Ile8]AngII and [Sar1Hcy3Cys5]AngII. Their ability to activate Gq, ERK1/2 and recruitment of β-arrestins (1 and 2) was evaluated. The analog [Sar1Hcy3Cys5]AngII appeared to activate the ERK1/2 pathway but not the G12 and β-arrestin pathways. These results suggest that constraining the movements of molecular determinants of a ligand by introducing cyclic structures can lead to a signaling bias by stabilizing different structures of the receptor.

Measuring and monitoring performance of employees is an integral part of management. The need for an effective appraisal system to manage and monitor the performance of employees is self-evident in this country. The appraisal system for office-based educators is informed by the Education Labour Relations Council (ELRC) Collective Agreement No. 3 of 2002, which was signed on 11 December 2002. This appraisal system was developed with good intensions of enhancing the performance of office-based educators, and the prime aim being to manage and improve service delivery at all levels in the system. The objective of the study is to investigate the perceptions of the office-based educators on the implementation of the appraisal system. Do office-based educators perceive it that way? As implementers and direct beneficiaries, are the good intensions of the policy yielded when its implementation is faced with realities and practicalities of the implementation of the policy? The vehicle I used to achieve the goal of the study was quantitative approach. Self administered questionnaires were completed by respondents. The questionnaire consisted of one open-ended question which gave the respondents sufficient room to voice their opinions without restrictions and the close-ended questions. The sample was the whole population of the office-based educators in the Dr Ruth S. Mompati district office, department of education, North West province. The total number is only 103 and it was manageable. The analysis of data is illustrated in the form of tables, graphs and brief discussions. A written undertaking to guarantee confidentiality to respondents was given and all respondents signed the undertaking to indicate their voluntary participation. The study is significant in that the findings and recommendations might inform policy custodians on the status and improvement of the quality of the implementation of the appraisal system. Copyright / Dissertation (MEd)--University of Pretoria, 2010. / Education Management and Policy Studies / unrestricted

I study movie studio profit maximization based on an optimization of a political language in the dialogues. I explore the flexibility with which a rational firm slants language of its movies in order to get closer either to a Democratic or a Republican customer. Using computational linguistics I construct vectors of phrase frequency distribution based on a text of almost a decade of U.S. Congress transcripts and 457 randomly chosen movie subtitles. In order to measure distance between the phrase vectors I use chi square statistics and its Monte Carlo approximation. I find no evidence of political slant in movies neither in a movie studio comparison nor for a time-varying comparison of movies in different years. In addition I construct a slant index covering level of political language in a movie. Using the index I find no evidence of impact of political language on movie revenues.

A mineração de dados é um processo computacionalmente caro, que se apoia no pré-processamento dos dados para aumentar a sua eficiência. As técnicas de redução de elementos do conjunto de dados, principalmente a amostragem de dados se destacam no pré-processamento. Os dados reais são caracterizados pela não uniformidade da distribuição, grande quantidade de atributos e presença de elementos considerados ruídos. Para esse tipo de dado, a amostragem uniforme, na qual cada elemento tem a mesma probabilidade de ser escolhido, é inefiiente. Os dados nos últimos anos, vem passando por transformações. Assim, não só o seu volume tem aumentado significantemente, mas também a maneira de como eles são representados. Os dados usualmente são divididos apenas em dados tradicionais (número e pequenas cadeias de caracteres) e dados complexos (imagens, cadeias de DNA, vídeos, etc). Entretanto, uma representação mais rica, na qual não só os elementos do conjunto são representados mas também a suas ligações, vem sendo amplamente utilizada. Esse novo tipo de dado, chamado rede complexa, fez surgir uma nova área de pesquisa chamada mineração de redes complexas ou de grafos, já que estes são utilizados na representação das redes complexas. Para esta nova área é necessário o desenvolvimento de técnicas que permitam a mineração de grandes redes complexas, isto é, redes com centenas de milhares de elementos(nós) e ligações(arestas). Esta tese teve como objetivo explorar a redução de elementos em conjuntos de dados chamados desbalanceados, isto é, que possuem agrupamentos ou classes de tamanhos bastantes distintos, e que também possuam alta quantidade de atributos e presença de ruídos. Além disso, esta tese também explora a mineração de redes complexas com a extração de padrões e propriedades e o desenvolvimento de algoritmos eficientes para a classificação das redes em reais e sintéticas. Também é proposto a mineração de redes complexas utilizando gerenciadores de base de dados para a mineração de cliques de tamanho 4 e 5 e a apresentação da extensão do coeficiente de clusterização / Data mining is an expensive computational process speeded up by data preprocessing. Data reduction techniques, as data sampling are useful during the data preprocessing. Real data are known for presenting non-uniform data distribution, a large amount of attributes and noise. For this type of data, uniform sampling, which selects elements with the same probability, is inefficient. Over the past years, the data available to mining have been changed. Not only have their volume increased but also data format. Data are usually divided into traditional (number and small chains of character) and complex (images, DNA, videos, etc). However, a rich representation, in which not only elements but also the connections among the elements have been used, is necessary. This new data type, which is called complex network and is usually modeled as a graph, has created a new research area, called graph mining or complex network mining, which requires the development of new mining techniques to allow mining large networks, that is, networks with hundreds of thousands of nodes and edges. The present thesis aims to explore the data reduction in unbalanced data, that is, data that have clusters with very different sizes, a large amount of attributes and noise. It also explores complex network mining with two basic findings: useful new patterns, which allow distinguishing real from synthetic networks and mining cliques of sizes 4 and 5 using database systems, discovering interesting power laws and presenting a new cluster coefficient formula