Honors thesis (English)--Oberlin College, 2009. / "April 2009" Includes bibliographical references (leaves 37-38)

Thesis (Ph. D.)--University of Washington, 1996. / Vita. Includes bibliographical references (leaves [108]-112).

Thesis (M.A.)--Trinity International University, 2001. / Abstract. Includes bibliographical references (leaves 116-125).

Thesis (M.A.)--Trinity International University, 2001. / Abstract. Includes bibliographical references (leaves 116-125).

Thesis (Ph. D.)--University of Wisconsin--Madison, 1975. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Bibliographical footnotes.

Nas últimas décadas, o grande avanço da ciência e tecnologia com suas invenções, novos materiais, equipamentos e métodos gerou a necessidade da criação de novos nomes, chamados aqui de termos, e alterações nos seus significados, para nomear adequadamente esses avanços, principalmente em áreas dinâmicas como a Ciência da Computação, a Genética e a Medicina. Dado que o desenvolvimento de repertórios terminológicos é um trabalho difícil quando realizado manualmente, lingüistas computacionais, lingüistas aplicados, tradutores, intérpretes, jornalistas científicos têm se interessado pela extração automática de terminologias (EAT) de textos. O crescimento explosivo de dados do tipo texto disponíveis na Web foi um fator contribuinte para a facilidade na construção de córpus eletrônicos de textos técnicos e científicos, propiciando a implementação de métodos de EAT. A EAT tem sido de grande interesse para todos os tipos de aplicações do Processamento de Línguas Naturais (PLN) que trabalham com domínios especializados e que, conseqüentemente, necessitam de um vocabulário especial. O objetivo desse projeto de mestrado foi avaliar métodos de EAT para o português do Brasil, ainda carente do tratamento automatizado para a criação de terminologias. Especificamente, foram implementados e avaliados métodos de EAT das abordagens estatística, lingüística e híbrida para unigramas, bigramas e trigramas a partir de um córpus de textos do domínio de Revestimentos Cerâmicos. Esses métodos empregam recursos simples como (a) uma stoplist para eliminar palavras como advérbios, (b) padrões sintáticos para os termos do domínio, por exemplo &ltsubstantivo adjetivo&gt, &ltsubstantivo preposição adjetivo&gt, levantados após a aplicação de um etiquetador Part-Of-Speech, (c) uma lista de expressões e palavras características de definições, descrições, classificações como 'definido(a)(s) como', 'caracterizado(a)', 'conhecido(a)(s) como', 'significa(m)', entre outras que são concentradoras de termos. As medidas estatísticas utilizadas nos métodos estatísticos e híbridos para indicar a relevância de termos no domínio são a informação mútua, o log-likelihood, o coeficiente dice e a freqüência. Os métodos propostos foram avaliados pelas medidas de precisão, revocação e medida F, utilizando uma lista de referência da área de Revestimentos Cerâmicos. Os melhores resultados da precisão são do método híbrido para unigramas (7%), bigramas (17%) e trigramas (26%), enquanto que a revocação é melhor nos métodos puramente lingüísticos tanto para unigramas (95%) como para bigramas (90%) e trigramas (100%). Os melhores valores da medida F foram dos métodos híbridos (11%, 17% e 33% para uni, bi e trigramas, respectivamente). Esses valores, embora tenham se apresentado os mais relevantes, foram bastante inferiores àqueles normalmente encontrados na literatura que trata da EAT, cujo desempenho obtido para essa tarefa fica em torno de 60%. Esses valores motivam a busca e implementação de métodos mais avançados para tratar o português, bem como a obtenção de recursos mais elaborados, a fim de encontrar resultados mais significantes para essa tarefa, facilitando, conseqüentemente o trabalho do especialista da área, que vai analisar os candidatos a termos extraídos pelos métodos automáticos, visto que é possível fornecer a ele informações mais precisas (poucas palavras da língua geral) e completas (uma maior quantidade de termos) sobre o córpus considerado. / During the last decades, the great advance in science and technology and their inventions, new materials, equipment and methods had as one result the necessity of creation of new names, called here terms, and alterations on their meanings, to name adequately these advances, mainly in areas as Computer Science, Genetics and Medicine. Considering that the development of terminological lists is an arduous work if manually executed, computational linguists, applied linguists, translators, interpreters and scientific journalists have been interested on automatic extraction of terminologies (AET) from texts. The sudden growing of data available on the Web was a contributing factor to facilitate the construction of electronic corpus of technical and scientific texts, providing implementation of AET methods. AET is very important for every sort of Natural Language Processing (NLP) applications that works on specialized domains and, consequently, needs special vocabulary. The purpose of this MS project was to evaluate AET methods for Brazilian Portuguese particularly, which is a language still in need of development of automatic treatment for terminology. Specifically, AET methods with statistic, linguistic and hybrid approaches were implemented and evaluated for unigrams, bigrams and trigrams for a corpus of texts in the domain of Ceramic Tiles. These methods use simple resources as (a) stoplist to eliminate words as adverbs, (b) syntactic patterns for terms from the domain, as, for instance, &ltsubstantive adjective&gt, &ltsubstantive preposition adjective&gt, considered after the application of a tagger Part-Of-Speech, (c) list of expressions and words typical of definitions, descriptions and classifications, like, for instance, 'defined as', 'characterized as', 'known as', 'that means', among others that concentrate terms. The statistic measures used by statistic and hybrid methods to indicate the terms relevance in the domain are mutual information, log-likelihood, dice coefficient, and frequency. The methods proposed were evaluated by precision, recall and F-measure, using a reference list in the area of Ceramic Tiles. The best results for precision are from the hybrid method for unigrams (7%), bigrams (17%) and trigrams (26%), while for recall the best results are from purely linguistic methods for unigrams (95%) as well as for bigrams (90%) and trigrams (100%). The best values for F-measure are from hybrid methods (11%, 17% and 33% for uni, bi and trigrams, respectively). These values, although presented as the most relevant ones, were quite inferior when compared to those commonly found in the literature concerned with AET, whose performance obtained for this task is around 60%. These values motivate the search and implementation of more advanced methods for Portuguese treatment, as well as the obtainment of more elaborated resources, in order to find more significant results for this task. In this way, the work of analysis of possible terms extracted by automatic methods done by the specialist of the area becomes much easier, since it is possible to provide him/her more precise (few word from general language) and complete (greater number of terms) information about the corpus under consideration.

Crude extracts of the seed kernels of the neem tree (Azadirachta indica) are widely used as plant protection products. The active ingredient (a.i.) of these extracts is azadirachtin A (aza A). aza A is a phytochemical (botanical) complex secondary metabolite which, with it is multiple toxic effects on insects, protects the plant against predation. Aza A is present in only low concentration in neem oil, but makes up 20-50% in the NSKEs extracted by polar solvents from the kernels. However, when used as foliar sprays it is rapidly destroyed by sunlight, and might be more effective if it is used systemically. Therefore the aim of the project was to extend previous work and to prepare a pelleted version of the main commercially-available neem-seed kernel extract, NeemAzal®-Technical (NAT) produced by Trifolio GmbH, in preparation for the expected registration of the product in the UK in 2011. It was first necessary to purify a quantity of aza A for quantification of the a.i. pelleted material and in soil and plants in the rest of the project. In achieving high purity (over 98%) aza A, reverse phase chromatographic methods were used, and mass spectrometery was used to confirm purity and identification. A final quantity of 6.2 mg of azadirachtin A was obtained from 4 gm of NAT, a yield of 0.15%. If aza A and the other neem terpenoids are to be used to plant protection, they must have a low phytotoxicity. Effect of NAT on the germination and its ensuing seedling development of two commercially important crops, sugar beet and cabbage was examined. NAT did have an inhibitory effect on seedling growth at 10-3 M aza A. In order to explore the inhibitory affect of aza A, the second part of the chapter was to examine effect of aza A on mitosis of onion root tips. The limonoids in concentration of 10-3 M adversely affected the mitotic activity of onion root tip cells. This could be failure of microtubules polymerisation into microtubules, or some other biochemical effect. From the findings in this part of the project, it can be concluded that only at a concentration of 10-3 M is aza A toxic to plant young seedlings, but in practice this is unlikely to be a significant problem. The first part of Chapter 4 of the project was to lay the foundations for the behaviour of aza A in soil environment in both powder form and in 2 types of granular formulations. The half-life of azadirachtin in soil from this work was found to be 1.6 days which is consistent with the previous reports. This short half-life of aza A may be problematic in use as a PPP. The short persistence might be overcome by formulating neem materials in granules to achieve environmental stability and biological efficacy of application. The granular formulations used in the project showed controlled release characteristics. The release of azadirachtin into the soil water was in fact delayed by encapsulating it in pellets. Systemic uptake of aza A by roots and subsequent presence in the vascular system of plants was assessed. Aza A was transported and was more stable in the leaf areas of cabbage and sugar beet plants than in the soil, as the half-life was found to be 9 days. The concentration of aza A in the leaf-water was less than 10% of the solution bathing the roots. The final part of the project, the application of the pelleted NSKE to protect cabbage, in both glass house and field conditions, demonstrated that neem products in pelleted formulations could be used as effective, systemically applied PPP to control pests of cabbage. In the field tests, the protective effect of the neem extract could be shown over a period of at least 5 weeks after addition of the pellets to the soil. In conclusion, the short soil half-life of the neem a.i., aza A, in PPP could be overcome by a pelleted formulation, the composition of which can delay release of the a.i. The technology allows protection of crops from soil-borne, as well as foliar sucking and biting pest damage by controlled release into the soil to allow uptake into plant vascular system.

In the field crops are subject to a wide variety of biotic and abiotic stresses. In order to manage crop protection effectively it is important that we understand not only the way plants respond to these stresses, but the way in which these responses interact. High levels of fertilisers and pesticides are often applied to maintain soil nutritional status and prevent disease in modern intensive farming systems. Potassium (K) is an essential element for plant growth and development, and is required for a wide variety of processes within the plant. These processes can be broadly divided into biophysical processes such as stomatal opening and cell extension and biochemical processes such as protein synthesis and enzyme activation. K starvation has been shown to lead to increased levels of the stress hormone jasmonate (JA) and related compounds in Arabidopsis thaliana plants which in turn modulates the plant’s defence against herbivorous insects and probably other pests or pathogens (Armengaud et al., 2004; Troufflard et al., 2010). In order for these results to be applicable to agriculture it is important to assess whether crop plants respond to K in a similar manner as the model plant. In this project the effect of K-deficiency on growth, metabolite concentrations, transcript levels and pathogen susceptibility of barley were investigated. Plants were grown in full-nutrient (control) or K-free hydroponic culture. The physiological, biochemical and transcriptional effects of K-deprivation were accessed over a time course of 20 days. Roots and shoots from plants grown in K-free nutrient solution had significantly lower K concentration than those grown in the control solution after 3 and 6 days respectively. A significant reduction in growth was seen as early as 6 days after K withdrawal. K-starvation led to a slight decrease in nitrogen metabolism, while hexose sugars strongly accumulated. By day 9 a significant increase in the expression of JA marker genes was seen in plants grown in K-free nutrient solution. Thus, despite possible differences in downstream events an induction of JA biosynthesis in response to K-deficiency occurs in both Arabidopsis and barley. Detached leaf segments were used to assess the effect of K-deficiency on infection of barley by two fungal pathogens with different strategies for nutrient acquisition. K-deficient barley plants were less susceptible to the biotroph Blumeria graminis f. f. sp. hordei (powdery mildew) and more susceptible to the hemi-biotroph Rhynchosporium secalis. Treatment of detached leaves with methyl-jasmonate (Me-JA) also led to less B. graminis infection, but had no effect on the R. secalis infection, indicating that JA increase in response to K-deficiency influences B. graminis but not R. secalis infection. The study therefore provides strong evidence that the effect of K-deficiency on pathogen susceptibility is determined by the JA-sensitivity of the pathogen.

The approach used in this study tested the application of an isotope dilution technique (ID) as a means of measuring the labile U(VI) and Th pools in soils. Uranium and Th lability and solubility were investigated for two sets of soils. The first set (Field soils) consisted of thirty seven soil samples representing five contrasting local ecosystems; the second dataset (BGSc) included 40 soils sub-sampled from the British Geological Survey (BGS) archive. Field soil pore water samples were taken from soil columns held at close to field capacity to measure U and Th solubility and speciation; the effects of time, temperature and reducing conditions on Th and U speciation were investigated. Soils were extracted with foursingle extractants: CH3COONH4, EDTA, 0.43 M HNO3 and TMAH to determine their ability to solubilize labile U and Th. Solubility of Th and U varied with soil characteristics, influenced by pH, DOC, DIC and phosphate concentrations. The Kd values for Th and U varied by 4 and 3 orders of magnitude respectively over the range of soils studied. The formation of soluble uranyl carbonate complexes give rise to a strong positive correlation between U and DIC concentrations in soil solutions. This was particularly clear under anaerobic conditions and also at high temperatures which encouraged microbial activity and high CO2partial pressures. Isotopically exchangeable 238U(VI) (the ‘E-value’, UE) in the soils studied varied from 2.7 to 39.1% of the total soil U content. On average, over all groups of soils, CH3COONH4, EDTA and TMAH underestimated E-value by factors of 13.7, 9.5 and 1.6, respectively, while extraction with 0.43M HNO3 overestimated E-value by only a factor of 1.04. Thus, on average across a range of soils, dilute nitric acid gave the best estimate of E-value compared to other extractants. Generally, E-values for U(VI) did not correspond consistently with any single chemical extraction procedure although the degree of correspondence was soil-dependent. Using UEand ThTMAH as input parameters in the geochemical speciation model WHAM-VII improved the prediction of U and Th solubility compared to using the total metal content orthe pools extractable by (other) single extraction methods. Finally, preliminary experiments confirmed the validity of ID for measuring labile soil Th without disturbance of soil-solution equilibrium.

pC221 belongs to a family of staphylococcal plasmids, including pT181, pS194, pC223, pUB112 and pCW7 . All possess open reading frames with 70-80% homology to the pC221 rapD gene. REP D has sequence specific topoisomerase activity at the pC221 origin (or iD) which is thought to be involved in replication initiation. DNase I footprinting has been carried out, showing that REP D binds to a region of oriD downstream of the nick site. The pattern of DNase I cleavage suggests that REP D contacts one face of the DNA helix, which may be bent around the protein. Extracts of S. aureus support incorporation of radioactive dNTPs into pC221 in the presence of REP D. Labelling with a[32P] dATP shows that replication initiates within the region containing oriD and proceeds in the direction expected for elongation of a 3' OH generated by nicking at oriD. With supercoiled DNA, REP D initiates replication of other members of this plasmid family in Vitro. However, with relaxed DNA, REP D is specific for oriD, suggesting that a change in the DNA, stabilised by supercoiling of the DNA or by binding of REP D, may be required for nicking. Of three inverted repeat sequences (ICRI, II & III) at the origin, ICRII has the greatest predicted hairpin stability and is almost totally conserved. Nicking takes place within the loop of this proposed hairpin. Disruption of base pairing within this hairpin has been investigated by mutagenesis of cloned oriD and using oligonucleotides based on the ICRII sequence. These experiments show that the 3' side of ICRII is more important for nicking than the 5' side. This is in agreement with footprinting data which shows that REP D binds the 3' side of ICRII, along with the whole of ICRIII. However, there is no evidence for hairpin formation at ICRII being required for nicking.