A liquid-to-air membrane energy exchanger (LAMEE) is a novel flat-plate membrane-based energy exchanger where heat and moisture transfer between air and solution streams occurs through a semi-permeable membrane. The LAMEE consists of many air and solution flow channels, each separated by a membrane. A small-scale single-panel LAMEE consists of a single pair of neighboring air and solution channels. This PhD thesis focuses on developing, testing and modeling the small-scale single-panel LAMEE, and investigating the similarity between the small-scale LAMEE and a full-scale LAMEE. This PhD thesis presents a methodology to investigate similarity between small-scale and full-scale energy exchangers. A single-panel energy exchanger test (SPEET) facility is developed and built to measure the performance of the small-scale single-panel LAMEE under different test conditions. Also, the small-scale LAMEE is numerically modeled by solving coupled heat and mass transfer equations for the air, solution and membrane of the LAMEE. The effects of membrane vapor diffusion resistance and enhanced air side convective heat transfer coefficient are numerically investigated. The numerical model of the small-scale LAMEE is validated with the experimental data for summer test conditions, and effectiveness values agree within ±4% in most cases. Moreover, the effects of different heat and mass transfer directions, and salt solution types and concentrations are experimentally and numerically investigated. The results show that the LAMEE effectiveness is strongly affected by the heat and mass transfer directions but negligibly affected by salt solution type and concentration. The solution-side effectiveness for liquid-to-air membrane energy exchangers is introduced in this thesis for the first time. The results show that the solution-side effectiveness should be used to evaluate the sensible and total effectiveness of LAMEE regenerators. Finally, the similarity between the small-scale and full-scale LAMEEs is investigated experimentally and numerically. The results show that the small-scale LAMEE effectiveness results can be used to predict the performance of a full-scale LAMEE within ±2% to ±4% in most cases.

While text Information Retrieval applications often focus on extracting semantic features to identify the topic of a document, and Music Information Research tends to deal with melodic, timbral or meta-tagged data of songs, useful information can be gained from surface-level features of musical texts as well. This is especially true for texts such as song lyrics and poetry, in which the sound and structure of the words is important. These types of lyrical verse usually contain regular and repetitive patterns, like the rhymes in rap lyrics or the meter in metrical poetry. The existence of such patterns is not always categorical, as there may be a degree to which they appear or apply in any sample of text. For example, rhymes in hip hop are often imperfect and vary in the degree to which their constituent parts differ. Although a definitive decision as to the existence of any such feature cannot always be made, large corpora of known examples can be used to train probabilistic models enumerating the likelihood of their appearance. In this thesis, we apply likelihood-based methods to identify and characterize patterns in lyrical verse. We use a probabilistic model of mishearing in music to resolve misheard lyric search queries. We then apply a probabilistic model of rhyme to detect imperfect and internal rhymes in rap lyrics and quantitatively characterize rappers' styles in their use. Finally, we compute likelihoods of prosodic stress in words to perform automated scansion of poetry and compare poets' usage of and adherence to meter. In these applications, we find that likelihood-based methods outperform simpler, rule-based models at finding and quantifying lyrical features in text.

Nacionalinių mokinių pasiekimų tyrimo metodologija yra nuolat plėtojama. Pasitarus su Švietimo plėtotės centro tyrimo skyriaus darbuotojais, buvo nuspręsta įsigilinti į klasterinės ir diskriminantinės analizės metodus, jų pritaikomumą nacionalinių tyrimų rezultatų analizei. Tai įtakojo šio darbo pasirinkimą ir tikslą - susipažinti su Klasterinės ir Diskriminantinės analizės metodais, pateikti ir aptarti jų taikymo pavyzdžius. Darbe išnagrinėti du daugiamačiai statistiniai metodai: klasterinė ir diskriminantinė analizė. Yra aptarti šių metodų teoriniai aspektai bei pateikti originalūs šių metodų pritaikymo mokinių pasiekimų tyrimams pavyzdžiai. Padarytos išvados apie šių metodų pritaikomumą nacionalinių mokinių pasiekimų tyrimo analizei atlikti. / National research of student marches are continually evolving. With Educational evolve center’s help I decided to analyse Cluster and Disckriminant analysis methods and how they can be used in National research analysis. This was a reason of my work purpose - to have a look at Cluster and Disckriminant analysis methods, to offer examples of practical use. There are two statistical multidimensional methods discussed in this work (Cluster and Disckriminant analysis). There are discussed theoretical side of these methods, also you can find original examples of practical use. In conclusion I discuss about practical use of these methods in national research of student marches.

<p>This study investigates the most important reasons behind the rst-year students' decision to select University of the Western Cape (UWC) as higher education institution.<br /> These reasons were organized into a few factors for easy interpretation. The data to be analyzed for this project is a subsection of the data collected during the orientation period of 2008. During the orientation week of 2008, the questionnaires were completed on a voluntary basis by new rst-year students. All questionnaires were anonymously completed and therefore the data does not contain any information that could be linked to any individual. For the purpose of this study, only the black African and coloured students were considered. The other racial groups were not analyzed due to too small sample sizes. Questionnaires with missing information on the reasons for selecting UWC were not&nbsp / nalyzed. We ended up with a sample of size 600. The data were statistically analyzed, using descriptive statistics, bivariate analyses, factor analysis, coefficient of congruence and bootstrap factor analysis. The results indicated that the most important reasons aecting students to choose UWC were identied as good academic reputation, family member's advice, UWC graduates are successful and UWC graduates get good jobs. The least important reasons were found to be not accepted anywhere, parents / family members graduated from UWC, recruited by UWC and wanted to study near to home. The results also indicated that there were significant differences among students according to population groups, parent's monthly income and grade 12 average. Factor analysis of 12 variables yielded three extracted factors upon which student decisions were based. Similarities of these three factors were tested, and a high similarity among demographic characteristics and grade 12 average were found. Additional analyses were conducted to measure the accuracy of factor analyses models constructed using Spearman and Polychoric correlation matrices. The results indicated that both correlation matrices were&nbsp / nbiased, with higher variance and higher loadings when the Polychoric correlation matrix was used to construct a factor analysis model for categorical data.</p>

Mémoire numérisé par la Division de la gestion de documents et des archives de l'Université de Montréal

Nearest-neighbor search is a very natural and universal problem in computer science. Often times, the problem size necessitates approximation. In this thesis, I present new paradigms for nearest-neighbor search (along with new algorithms and theory in these paradigms) that make nearest-neighbor search more usable and accurate. First, I consider a new notion of search error, the rank error, for an approximate neighbor candidate. Rank error corresponds to the number of possible candidates which are better than the approximate neighbor candidate. I motivate this notion of error and present new efficient algorithms that return approximate neighbors with rank error no more than a user specified amount. Then I focus on approximate search in a scenario where the user does not specify the tolerable search error (error constraint); instead the user specifies the amount of time available for search (time constraint). After differentiating between these two scenarios, I present some simple algorithms for time constrained search with provable performance guarantees. I use this theory to motivate a new space-partitioning data structure, the max-margin tree, for improved search performance in the time constrained setting. Finally, I consider the scenario where we do not require our objects to have an explicit fixed-length representation (vector data). This allows us to search with a large class of objects which include images, documents, graphs, strings, time series and natural language. For nearest-neighbor search in this general setting, I present a provably fast novel exact search algorithm. I also discuss the empirical performance of all the presented algorithms on real data.

Dans les turbomachines conventionnelles, l’estimation des performances (rendement, puissance et rapport de pression) se fait en général en admettant l’adiabaticité de l’écoulement. Mais, de nombreuses études ayant montré l’influence négative des échanges thermiques internes et externes sur les performances des petites turbomachines dans les faibles charges et aux bas régimes, cette hypothèse ne peut plus être recevable. L’objectif principal de cette thèse est de contribuer à lever l’hypothèse d’adiabaticité.Une étude préalable de l’état de l’art a permis de relever les différents types de transferts thermiques dans les turbomachines et de circonscrire notre étude.Puis, une analyse exergétique généralisée, ayant pour but la prise en compte des deux principes de la thermodynamique, a été effectuée et l’évolution de l’indice de performance caractérisant le niveau d’énergie récupérable en fonction des échanges thermiques est étudiée.Les performances des turbomachines à fluide compressible sont généralement représentées sous forme graphique dans des systèmes de coordonnées adimensionnelles établies avec l’hypothèse d’adiabaticité. Ces cartographies couramment utilisées par les exploitants et constructeurs ne conviennent pas aux machines fonctionnant avec transferts thermiques. L’étude de la similitude des turbomachines thermiques à fluide compressible présentée dans ce travail, propose de nouvelles coordonnées adimensionnelles pouvant être utilisées aussi bien en adiabatique que dans les écoulements avec transferts thermiques.Enfin, nous proposons un protocole de mesures et un modèle numérique pour l’évaluation des transferts thermiques dans un turbocompresseur.Certains résultats obtenus montrent que les performances calculées avec l’hypothèse d’adiabaticité de l’écoulement du fluide sont surestimées. Les nouvelles lois de la similitude proposées généralisent le théorème de Rateau au fluide compressible fonctionnant dans n’importe quelle condition et permettent de calculer les échanges thermiques à chaud à partir des résultats d’essai à froid. Une donnée supplémentaire (température de refoulement) est néanmoins nécessaire pour la prédiction complète des performances et des échanges thermiques.Le modèle numérique de calcul des échanges thermiques proposé donne des résultats en accord avec ceux attendus, mais nécessite des données réelles issues de mesure sur banc pour une validation complète. / In the conventional turbomachines, calculations are done assuming adiabatic flow. But, the negative influence of external and internal heat exchange on the performance of small turbomachines at low loads and low speeds have been shown by many studies in the literature. Then, this assumption is no longer admissible. The main objective of this thesis is to help remove the assumption of adiabaticity.A study of the state of art has identified the different kinds of heat transfer and defined the limits of our investigations.Afterwards, a generalized exergy analysis whose main goal is to take into account the two principles of thermodynamics has been made and the variation of exergy performance versus heat transfer has been studied.The maps currently used are made with the assumption of adiabaticity. The laws of similarity in turbomachines working with compressible fluid studied propose new dimensionless coordinates that can be used in any operating condition (adiabatic or not).Finally, we present a measurement protocol and a numerical model for calculating heat transfer in a turbocharger.Some results from our work indicate that the performance of thermal turbomachinery announced regardless of thermal heat exchanges are found to be overestimated.The new laws of similarity proposed generalize the Rateau’s theorem to compressible fluid flow in any operating condition and can be used to calculate heat transfer from adiabatic test results. Supplementary information is still required for the complete prediction of performance and heat transfer.The numerical model for calculating heat transfer gives some results that are in agreement with those expected. But actual data from test bench are required for complete validation.

Drug repositioning applies existing drugs to new disease indications. A prerequisite for drug repurposing is drug promiscuity - a drug's ability to bind to several targets, possibly leading to side effects on the other hand. One reason for drug promiscuity is binding site similarity between (otherwise unrelated) proteins. In this thesis, a new algorithm for remote binding site similarity assessment and its application to the whole of the Protein Data Bank (PDB) is presented, forming the base for off-target identification and drug repositioning. The present thesis contributes to a long-standing debate on the reasons for drug promiscuity, being one of the pioneer studies investigating these from a protein structural point of view. Except for a small influence of flexibility, the analysis of all promiscuous drugs in the PDB revealed that drug properties are of minor importance. However, a strong correlation between promiscuity and binding site similarity of protein targets is found (r = 0.81), suggesting binding site similarity as the main reason for drug promiscuity. For 71 % of the promiscuous drugs at least one pair of their targets' binding sites is similar and for 18 % all are similar. In order to overcome issues in detection of remotely similar binding sites, a score for binding site similarity is developed: LigandRMSD measures the similarity of the aligned ligands and uncovers remote local similarities in proteins. It can be applied to arbitrary binding site alignments and also works on distinct ligands on a structural proteome scale. To answer the question on which other targets might be hit when targeting a particular protein, an all-to-all binding site alignment of 32,202 protein structures is analyzed. Of the hundreds of million possible protein pairs, 0.27 % were found to have similar binding sites. Extrapolating to the human proteome, for one human protein are 54 proteins with a similar binding site expected on average. Clearly, this is in contrast to the one drug-one target paradigm in drug development. Based on these data, disadvantageous off-targets can be uncovered and drug-repositioning candidates inferred. The enormous potential is demonstrated with the example of Viagra, proposing it for repositioning to Alzheimer's disease and prostate cancer. The findings in this thesis question the established single-target dogma in drug discovery. Drugs are triggered to modulate multiple targets simultaneously by the widespread binding site similarity. With the presented pipeline, drug targets can be reliably predicted: Starting from a target protein, additional targets are predicted based on binding site similarity and prioritized according to the resulting ligand structural overlap. Identifying drug targets helps to understand severe side effects and opens the door for drug repositioning.

Unsupervised information extraction in open domain gains more and more importance recently by loosening the constraints on the strict definition of the extracted information and allowing to design more open information extraction systems. In this new domain of unsupervised information extraction, this thesis focuses on the tasks of extraction and clustering of relations between entities at a large scale. The objective of relation extraction is to discover unknown relations from texts. A relation prototype is first defined, with which candidates of relation instances are initially extracted with a minimal criterion. To guarantee the validity of the extracted relation instances, a two-step filtering procedures is applied: the first step with filtering heuristics to remove efficiently large amount of false relations and the second step with statistical models to refine the relation candidate selection. The objective of relation clustering is to organize extracted relation instances into clusters so that their relation types can be characterized by the formed clusters and a synthetic view can be offered to end-users. A multi-level clustering procedure is design, which allows to take into account the massive data and diverse linguistic phenomena at the same time. First, the basic clustering groups similar relation instances by their linguistic expressions using only simple similarity measures on a bag-of-word representation for relation instances to form high-homogeneous basic clusters. Second, the semantic clustering aims at grouping basic clusters whose relation instances share the same semantic meaning, dealing with more particularly phenomena such as synonymy or more complex paraphrase. Different similarities measures, either based on resources such as WordNet or distributional thesaurus, at the level of words, relation instances and basic clusters are analyzed. Moreover, a topic-based relation clustering is proposed to consider thematic information in relation clustering so that more precise semantic clusters can be formed. Finally, the thesis also tackles the problem of clustering evaluation in the context of unsupervised information extraction, using both internal and external measures. For the evaluations with external measures, an interactive and efficient way of building reference of relation clusters proposed. The application of this method on a newspaper corpus results in a large reference, based on which different clustering methods are evaluated.

Unsupervised information extraction in open domain gains more and more importance recently by loosening the constraints on the strict definition of the extracted information and allowing to design more open information extraction systems. In this new domain of unsupervised information extraction, this thesis focuses on the tasks of extraction and clustering of relations between entities at a large scale. The objective of relation extraction is to discover unknown relations from texts. A relation prototype is first defined, with which candidates of relation instances are initially extracted with a minimal criterion. To guarantee the validity of the extracted relation instances, a two-step filtering procedures is applied: the first step with filtering heuristics to remove efficiently large amount of false relations and the second step with statistical models to refine the relation candidate selection. The objective of relation clustering is to organize extracted relation instances into clusters so that their relation types can be characterized by the formed clusters and a synthetic view can be offered to end-users. A multi-level clustering procedure is design, which allows to take into account the massive data and diverse linguistic phenomena at the same time. First, the basic clustering groups similar relation instances by their linguistic expressions using only simple similarity measures on a bag-of-word representation for relation instances to form high-homogeneous basic clusters. Second, the semantic clustering aims at grouping basic clusters whose relation instances share the same semantic meaning, dealing with more particularly phenomena such as synonymy or more complex paraphrase. Different similarities measures, either based on resources such as WordNet or distributional thesaurus, at the level of words, relation instances and basic clusters are analyzed. Moreover, a topic-based relation clustering is proposed to consider thematic information in relation clustering so that more precise semantic clusters can be formed. Finally, the thesis also tackles the problem of clustering evaluation in the context of unsupervised information extraction, using both internal and external measures. For the evaluations with external measures, an interactive and efficient way of building reference of relation clusters proposed. The application of this method on a newspaper corpus results in a large reference, based on which different clustering methods are evaluated.