Using interaction data for improving the offline and online evaluation of search engines

This thesis investigates how web search evaluation can be improved using historical interaction data. Modern search engines combine offline and online evaluation approaches in a sequence of steps that a tested change needs to pass through to be accepted as an improvement and subsequently deployed. We refer to such a sequence of steps as an evaluation pipeline. In this thesis, we consider the evaluation pipeline to contain three sequential steps: an offline evaluation step, an online evaluation scheduling step, and an online evaluation step. In this thesis we show that historical user interaction data can aid in improving the accuracy or efficiency of each of the steps of the web search evaluation pipeline. As a result of these improvements, the overall efficiency of the entire evaluation pipeline is increased. Firstly, we investigate how user interaction data can be used to build accurate offline evaluation methods for query auto-completion mechanisms. We propose a family of offline evaluation metrics for query auto-completion that represents the effort the user has to spend in order to submit their query. The parameters of our proposed metrics are trained against a set of user interactions recorded in the search engine’s query logs. From our experimental study, we observe that our proposed metrics are significantly more correlated with an online user satisfaction indicator than the metrics proposed in the existing literature. Hence, fewer changes will pass the offline evaluation step to be rejected after the online evaluation step. As a result, this would allow us to achieve a higher efficiency of the entire evaluation pipeline. Secondly, we state the problem of the optimised scheduling of online experiments. We tackle this problem by considering a greedy scheduler that prioritises the evaluation queue according to the predicted likelihood of success of a particular experiment. This predictor is trained on a set of online experiments, and uses a diverse set of features to represent an online experiment. Our study demonstrates that a higher number of successful experiments per unit of time can be achieved by deploying such a scheduler on the second step of the evaluation pipeline. Consequently, we argue that the efficiency of the evaluation pipeline can be increased. Next, to improve the efficiency of the online evaluation step, we propose the Generalised Team Draft interleaving framework. Generalised Team Draft considers both the interleaving policy (how often a particular combination of results is shown) and click scoring (how important each click is) as parameters in a data-driven optimisation of the interleaving sensitivity. Further, Generalised Team Draft is applicable beyond domains with a list-based representation of results, i.e. in domains with a grid-based representation, such as image search. Our study using datasets of interleaving experiments performed both in document and image search domains demonstrates that Generalised Team Draft achieves the highest sensitivity. A higher sensitivity indicates that the interleaving experiments can be deployed for a shorter period of time or use a smaller sample of users. Importantly, Generalised Team Draft optimises the interleaving parameters w.r.t. historical interaction data recorded in the interleaving experiments. Finally, we propose to apply the sequential testing methods to reduce the mean deployment time for the interleaving experiments. We adapt two sequential tests for the interleaving experimentation. We demonstrate that one can achieve a significant decrease in experiment duration by using such sequential testing methods. The highest efficiency is achieved by the sequential tests that adjust their stopping thresholds using historical interaction data recorded in diagnostic experiments. Our further experimental study demonstrates that cumulative gains in the online experimentation efficiency can be achieved by combining the interleaving sensitivity optimisation approaches, including Generalised Team Draft, and the sequential testing approaches. Overall, the central contributions of this thesis are the proposed approaches to improve the accuracy or efficiency of the steps of the evaluation pipeline: the offline evaluation frameworks for the query auto-completion, an approach for the optimised scheduling of online experiments, a general framework for the efficient online interleaving evaluation, and a sequential testing approach for the online search evaluation. The experiments in this thesis are based on massive real-life datasets obtained from Yandex, a leading commercial search engine. These experiments demonstrate the potential of the proposed approaches to improve the efficiency of the evaluation pipeline.

Tributos indiretos na cadeia produtiva do camarão

The state of Rio Grande do Norte counts with a relevant potential in the shrimp farming supply chain. In the larviculture step the state responds for more than half of the national production. In the farming step it is the second largest producer. In the industrial step, its industries have almost 40% of the shrimp processing capacity of the northeast of Brazil. However, this country has the highest tax rate comparing with the main shrimp producer countries. Considering the influence of taxes in the competition among companies, the main goal of this research is to analyze the impact of indirect taxes in the above steps of the supply chain. To achieve it, it will be used the data of the 2011 Census of the Shrimp Farming and it will be applied the Herfindahl-Hirschman Index to identify the market form of those steps. In order to contribute with the characterization of the supply chain, CEO´s of farms and industries will be interviewed. The price-elasticity of the shrimp larvae, the in natura shrimp and the processed shrimp will be analyzed in order to verify the possibility that each one of those three steps has to pass-through the onus of the end of benefit over the ICMS. The data analysis shows that the larviculture step functions as a duopoly and, facing the end of that benefit, it will be able to pass-through most its onus to the farming step. On the other hand, this step functions similar to a perfect competing market, which diminishes its capacity to pass-through that onus to the processing step. This step operates as oligopoly with a lower concentration than the larviculture step but, due to the fact that it faces an oligopsony, it will end up assuming most of that onus, which will cause a decrease in the amount of processed shrimp. It is concluded that the end of that benefit would impact negatively, in this state, the supply chain at all, but mainly the farming and the industrial steps

INCORPORACIÓN DE NUEVAS HERRAMIENTAS DE PRONÓSTICO EN LA COMBINACIÓN DE PROYECCIONES DE INFLACIÓN

Esta nota tiene como objetivo incorporar en la proyección combinada de inflación el pronóstico de la tasa de inflación proveniente del Modelo Pass-Through. Los otros modelos individuales utilizados en la combinación de inflación corresponden a dos modelos autorregresivos (ARMA e Ingenuo) y dos modelos de vectores autorregrevios (VAR lineal y no lineal).Con respecto a las ponderaciones utilizadas actualmente para realizar la proyección de inflación se observa: i) una pérdida de ponderación para el modelo ingenuo; ii) un aumento en el peso asignado al modelo de petróleo; y iii) prácticamente la exclusión del modelo ARMA y VAR.. No obstante el cambio en la ponderación, la incorporación del modelo Pass Trough no mejoró la precisión de las proyecciones de inflación, evaluada por medio de la raíz del error cuadrático medio (RECM).Dados los resultados anteriores se sugiere el empleo de la nueva versión de combinación en las futuras proyecciones de inflación por cuanto: a) el modelo ingenuo pierde participación, solventando una de las principales limitaciones de la versión anterior, b) se incorpora un modelo con mayor contenido teórico-económico para explicar la inflación, c) se excluyen modelos de tipo autorregresivo y de vectores autorregresivos, los cuales se aduce no tienen como respaldo un modelo económico y d) no desmejora la precisión de la proyección, medida por la RECM.Así, con las nuevas ponderaciones e información a agosto la proyección combinada de la tasa de inflación a diciembre del 2003 es 8,9%, para diciembre del 2004 es 8,3% y en 24 meses es 8,0%. Estos resultados no difieren en gran medida de los reportados en el informe DIE-DCS-68-2003-IT, en el cual la inflación interanual se proyectaba en 8,8% para diciembre 2003 y 2004 y en 8,7% en 24 meses.

Relatório de estágio de intervenção comunitária

O presente relatório, é baseado no estágio de intervenção comunitária realizado na Escola Secundária Mouzinho da Silveira, em Portalegre. Teve por base um diagnóstico de situação nomeadamente sobre “Hábitos de Vida Saudáveis dos Adolescentes” e teve o seu foco de atenção em termos de população alvo, nos alunos adolescentes do 7º e 8º Ano, da escola anteriormente referida. O estudo efetuado utilizou como método a Metodologia Quantitativa. Tratando-se de um estudo do tipo Descritivo, Exploratório e Transversal, os dados foram obtidos através da aplicação de um questionário estruturado sobre as temáticas a serem abordadas. Este relatório tem como principal finalidade descrever as atividades desenvolvidas e as competências adquiridas como enfermeira especialista durante o estágio de intervenção comunitária e refletir sobre a aprendizagem desenvolvida bem como os progressos obtidos durante a sua prática. Teve ainda o objetivo de caracterizar o processo de Planeamento em Saúde, abordando as diferentes etapas ultrapassadas durante este projeto de intervenção. Este relatório permitiu ainda a abordagem de temáticas fundamentais neste tipo de intervenção tais como a promoção da saúde e a educação para a saúde, sua importância e suas vantagens quando se trata de intervir na comunidade melhorando os seus níveis de saúde e bem-estar. Através das conclusões obtidas e aqui referenciadas, foi possível uma avaliação das sessões realizadas bem como a sua eficácia

Ecogenômica de archaea e monitoramento de comunidades de procariotos redutores de sulfato: aplicações na indústria de petróleo e gás

The human activities responsible for the ambient degradation in the modern world are diverse. The industrial activities are preponderant in the question of the impact consequences for brazilian ecosystems. Amongst the human activities, the petroliferous industry in operation in Potiguar Petroliferous Basin (PPB) displays the constant risk of ambient impacts in the integrant cities, not only for the human populations and the environment, but also it reaches the native microorganisms of Caatinga ground and in the mangrove sediment. Not hindering, the elaboration of strategies of bioremediation for impacted areas pass through the knowledge of microbiota and its relations with the environment. Moreover, in the microorganism groups associated to oil, are emphasized the sulfate-reducing prokaryotes (SRP) that, in its anaerobic metabolism, these organisms participate of the sulfate reduction, discharging H2S, causing ambient risks and causing the corrosion of surfaces, as pipelines and tanks, resulting in damages for the industry. Some ancestries of PRS integrate the Archaea domain, group of microorganisms whose sequenced genomes present predominance of extremophilic adaptations, including surrounding with oil presence. This work has two correlated objectives: i) the detection and monitoring of the gene dsrB, gift in sulfate-reducing prokaryotes, through DGGE analysis in samples of mDNA of a mangrove sediment and semiarid soil, both in the BPP; ii) to relate genomic characteristics to the ecological aspects of Archaea through in silico studies, standing out the importance to the oil and gas industry. The results of the first work suggest that the petrodegraders communities of SRP persist after the contamination with oil in mangrove sediment and in semiarid soil. Comparing the populations of both sites, it reveals that there are variations in the size and composition during one year of experiments. In the second work, functional and structural factors are the probable cause to the pressure in maintenance of the conservation of the sequences in the multiple copies of the 16S rDNA gene. Is verified also the discrepancy established between total content GC and content GC of the same gene. Such results relating ribosomal genes and the ambient factors are important for metagenomic evaluations using PCR-DGGE. The knowledge of microbiota associated to the oil can contribute for a better destination of resources by the petroliferous industry and the development of bioremediation strategies. Likewise, search to lead to the best agreement of the performance of native microbiota in biogeochemical cycles in Potiguar Petroliferous Basin ecosystem

Arsenic Geochemitry in Warm Spring Ponds: New Field and Experimental Results

Silver Bow Creek (SBC) flows into the Warm Springs Ponds Operable Unit (WSPOU), where various containment cells are used to precipitate copper and other metals (e.g., Cd, Cu, Mn, Pb, Zn). Lime is added seasonally to increase the pH and assist in removal of metals from the water column. Although the WSPOU is effective at removing copper and other cationic trace metals, concentrations of dissolved arsenic exiting the facility are often above the site specific standard, 20 20 ug/L, during low-flow periods each summer and fall. This thesis is a continuation of arsenic geochemistry studies by Montana Tech in the WSPOU. Field work focused on Pond 3, the largest and first in the series of treatment ponds. Shallow groundwater was sampled from 8 PVC piezometers located near the south end of Pond 3. Three sediment pore-water diffusion samplers (“peepers”) were also deployed at the south end of Pond 3 to examine vertical gradients in chemistry in the top 25 cm of the pond sediment. In general, the pH and Eh values of the shallow groundwater and sediment pore-water were less than in the pond water. Concentrations of arsenic were generally higher in subsurface water, and tended to pass through a maximum (up to 530 g/L) about 10 cm below the sediment-water interface. In the peeper cells, there was a strong positive correlation between dissolved As and dissolved Fe, and an inverse correlation with sulfate. Therefore, the zone of arsenic release corresponds to a zone of bacterial Fe and sulfate reduction in the shallow, organic-rich sediment. Redox speciation of arsenic shows that arsenate (As(V)) is dominant in the pond, and arsenite (As(III)) is dominant in the subsurface water. A series of laboratory experiments with pH adjustment were completed using SBC water collected near the inlet to the WSPOU as well as water and shallow sediment collected from Pond 3. Water ± sediment mesocosms were set up in 1-L Nalgene bottles (closed system) or a 20-L aquarium (open system), both with continuous stirring. The pH of the mesocosm was adjusted by addition of NaOH or HNO3 acid. The closed system provided better pH control since the water was not in contact with the atmosphere, which prevented exchange of carbon dioxide. In both the closed and open systems, dissolved arsenic concentrations either decreased or stayed roughly the same with increase in pH to values > 11. Therefore, the release of dissolved As into the treatment ponds in low-flow periods is not due to changes in pH alone. All of these results support the hypothesis that the arsenic release in WSPOU is linked to microbial reduction of ferric oxide minerals in the organic-rich sediment. Upwards diffusion of dissolved As from the sediment pore-water into the pond water is the most likely explanation for the increase in As concentration of the WSPOU in low-flow periods.

Mediação Penal e Justiça Restaurativa

Projeto de Graduação apresentado à Universidade Fernando Pessoa como parte dos requisitos para obtenção do grau de licenciada em Criminologia

ANALISANDO A RELAÇÃO DOS VOLUMES DE EXPORTAÇÃO E DE IMPORTAÇÃO, PIB, TAXAS DE CÂMBIO E INFLAÇÃO NO PERÍODO DE 2004 A 2014.

Esta dissertação visa deslumbrar uma análise macroeconômica do Brasil, especialmente no que se refere à relação dos índices mensais dos volumes das exportações e das importações com os volumes mensais do PIB, da Taxa SELIC e as Taxas de Câmbio, conforme dados coletados no período de janeiro de 2004 a dezembro de 2014, através de pesquisa literária referente aos históricos sobre cada conceito envolvido no âmbito da macroeconomia das varáveis estudadas. Foi realizado um estudo de caso embasado em dados de sites governamentais, no período delimitado, empregando-se o método de regressão linear, com base na Teoria da correlação de Pearson, demonstrando os resultados obtidos no período do estudo para as varáveis estudadas. Desta maneira, conseguiu-se estudar e analisar como as variáveis dependentes (resposta): volume das exportações e volume das importações estão relacionadas com as varáveis independentes (explicativas): PIB, Taxa Selic e taxa de Câmbio. Os resultados apurados no presente estudo permitem identificar que existe correlação moderada e negativa, quando analisadas a Taxa Selic e a Taxa de Câmbio com os volumes das exportações e das importações, enquanto o PIB apresenta correlação forte e positiva na análise com os volumes das exportações e das importações

ANALISANDO A RELAÇÃO DOS VOLUMES DE EXPORTAÇÃO E DE IMPORTAÇÃO, PIB, TAXAS DE CÂMBIO E INFLAÇÃO NO PERÍODO DE 2004 A 2014.

Esta dissertação visa deslumbrar uma análise macroeconômica do Brasil, especialmente no que se refere à relação dos índices mensais dos volumes das exportações e das importações com os volumes mensais do PIB, da Taxa SELIC e as Taxas de Câmbio, conforme dados coletados no período de janeiro de 2004 a dezembro de 2014, através de pesquisa literária referente aos históricos sobre cada conceito envolvido no âmbito da macroeconomia das varáveis estudadas. Foi realizado um estudo de caso embasado em dados de sites governamentais, no período delimitado, empregando-se o método de regressão linear, com base na Teoria da correlação de Pearson, demonstrando os resultados obtidos no período do estudo para as varáveis estudadas. Desta maneira, conseguiu-se estudar e analisar como as variáveis dependentes (resposta): volume das exportações e volume das importações estão relacionadas com as varáveis independentes (explicativas): PIB, Taxa Selic e taxa de Câmbio. Os resultados apurados no presente estudo permitem identificar que existe correlação moderada e negativa, quando analisadas a Taxa Selic e a Taxa de Câmbio com os volumes das exportações e das importações, enquanto o PIB apresenta correlação forte e positiva na análise com os volumes das exportações e das importações. Palavras-

Juros, câmbio e o sistema de metas de inflação no Brasil

Interest rate, exchange rate and the system of inflation target in Brazil. In the consensus view of the Brazilian system of inflation targeting, the core of inflation is due to demand shocks; the rate of interest is set to control demand; and some variation in the exchange rate happens as "collateral damage". In this note we argue that in reality core inflation comes from cost push; the interest rate affects the exchange rate; changes in the exchange rate affect costs and prices; it is the effect of interest rates on demand that is the "collateral damage" and that the long run anchor of the system is low average real wage rigidity.

Metas de inflação, taxa de câmbio real e crise externa num modelo kaleckiano

Inflation target, real exchange rate and external crisis in a Kaleckian model. Which role should the real exchange rate play in an inflation target regime? In this paper this point is discussed from the point of view of the conditions required for avoiding an external crisis. With this objective, a dynamic Kaleckian model is presented focusing on the stability of the external debt to capital ratio. The main conclusion is that policy makers should monitor closely the evolution of the real exchange rate in order to make compatible the inflation target regime with external stability.

Perception of speech through a 500 Hz low-pass filter: sensory aids and communication strategies

This paper reviews a study to investigate how a hearing impaired person can learn to discriminate speech distorted by a low pass filter in a sensory aid.

Perception of speech through a 500 Hz low-pass filter: sensory aids and communication strategies

This paper reviews a study to investigate how a hearing impaired person can learn to discriminate speech distorted by a low pass filter in a sensory aid.

Volatilidade da taxa de câmbio em economias com altas taxas de inflação: um estudo econométrico do Brasil e a Polônia

A dissertação investiga as séries de taxa de câmbio da Polônia do Brasil. elevada volatilidade da série polonesa não explicada pelas variáveis utilizadas como fundamentos. No caso brasileiro, isto verdade para modelo com variáveis do setor real da economia Detectou-se co-integração entre as variáveis do setor financeiro, que levanta hipótese da existência de um processo de aprendizagem induzido pelo longo convívio com altas taxas de inflação e indexação generalizada da economia brasileira. volatilidade da taxa de cambio bastante impactada por intervenções drásticas na economia, concluindo-se que adoçáo deste tipo de medida atua como fator de incremento da volatilidade da taxa de câmbio.

Regime político e participação do legislativo na determinação da taxa de câmbio de países que conviveram com elevada inflação e persistentes déficits em transações correntes: os casos de Argentina, Bolívia, Brasil, México e Peru entre 1970 e 1989

O objetivo deste artigo é verificar a influência de variáveis políticas na determinação da taxa de câmbio em quatro países latino-americanos que conviveram com elevada inflação e déficit em Transações Correntes nas décadas de setenta e oitenta. Estudos empíricos já haviam demonstrado a influência das eleições. Nenhum, porém, havia incorporado a estrutura de decisão do Executivo e Legislativo neste processo. Só foi possível incorporar o regime político (Autoritário/Democrático) e a divisão de poder no Legislativo de todos os países num modelo standard de taxa de câmbio porque utilizamos a técnica de painel. Obtivemos os seguintes resultados: países classificados como Autoritários apresentaram uma taxa de câmbio mais valorizada e Legislativos mais fragmentados apresentaram uma taxa de câmbio mais desvalorizada. Vimos este último resultado com desconfiança uma vez que, entre os países da amostra, o regime Autoritário era, em alguns casos, uma ditadura militar e o Legislativo pouco intervia nas decisões. Interagimos o regime político com fragmentação e percebemos que o efeito da classificação do regime predomina. No caso, se existir um regime Autoritário, o câmbio resultante da interação ainda será valorizado. A divisão de poder no Legislativo apenas provoca uma redução no impacto da valorização.