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.

Geological Characterisation of Depleted Oil and Gas Reservoirs for Carbon Sequestration Potentials in a Field in the Niger Delta, Nigeria

The injectivity, containment and storage capacity of sandstone reservoirs in a field in the Coastal Swamp depobelt of the onshore eastern Niger Delta were evaluated using wireline logs and seismic data to assess their potentials for carbon dioxide storage and geosequestration. The reservoir formation consists of multilayered alternating beds of sandstone and shale cap rocks. Active seismicity and fracturing intensity are low and growth faults provide the reservoir sealing mechanisms. Three reservoirs were delineated at depths between 3319 m and 3539 m which will keep injected CO2 in a supercritical state. The reservoir depth of at least 800 m, porosity and permeability of more than 10 percent and 20 mD, and a caprock thickness of at least 10 m, in addition to geothermal gradients of 13.46 to 33.66 ºC /km are the ideal conditions for the efficacy of storage. Comparison of the derived reservoir and seal properties such as porosity, permeability, thickness and depth with the minimum recommended site selection criteria shows that the reservoirs are potential candidates for carbon geosequestration with a total theoretical storage capacity of 147MM tons.

Sémaphore, suivi de services opérationnels

L’unité IDM fournit des services opérationnels utilisés par différents utilisateurs des scientifiques, des partenaires mais également le grand public. L’unité IDM souhaite mieux connaitre ses utilisateurs et évaluer la performance de ses services, afin : . D’améliorer le service rendu . D’effectuer du reporting mensuel/annuel . De répondre aux engagements de service (SLA) Pour cela, il faut utiliser différentes sources d’informations : Monitoring : permet de surveiller les services opérationnels et de connaitre l’état actuel et déterminer le taux de disponibilité . Logs d’activité o Logs serveurs (http, ftp, … ) : permet de connaître l’activité réelle des utilisateurs o Logs applicatifs : permet de récupérer des informations plus précise sur les différentes briques logicielles des services opérationnels L’objectif IDM, à travers cette alternance est de mettre en place une infrastructure commune à l’ensemble des systèmes d’informations.

I. Nickel-Exchanged Zincosilicate Catalysts for the Oligomerization of Propylene and II. Organic SDA-Free Catalysts for the Methanol-to-Olefins Reaction

Nickel-containing catalysts are developed to oligomerize light olefins. Two nickel-containing zincosilicates (Ni-CIT-6 and Ni-Zn-MCM-41) and two nickel-containing aluminosilicates (Ni-HiAl-BEA and Ni-USY) are synthesized as catalysts to oligomerize propylene into C_3n (C₆ and C₉) products. All catalysts oligomerize propylene, with the zincosilicates demonstrating higher average selectivities to C_3n products, likely due to the reduced acidity of the Zn heteroatom.

To test whether light alkanes can be incorporated into this oligomerization reaction, a supported homogeneous catalyst is combined with Ni-containing zincosilicates. The homogeneous catalyst is included to provide dehydrogenation/hydrogenation functions. When this tandem catalyst system is evaluated using a propylene/n-butane feed, no significant integration of alkanes are observed.

Ni-containing zincosilicates are reacted with 1-butene and an equimolar propylene/1-butene mixture to study other olefinic feeds. Further, other divalent metal cations such as Mn²⁺, Co²⁺, Cu²⁺, and Zn²⁺ are exchanged onto CIT-6 samples to investigate stability and potential use for other reactions. Co-CIT-6 oligomerizes propylene, albeit less effectively than Ni-CIT-6. The other M-CIT-6 samples, while not able to oligomerize light olefins, may be useful for other reactions, such as deNO_x.

Molecular sieves are synthesized, characterized, and used to catalyze the methanol-to-olefins (MTO) reaction. The Al concentration in SSZ-13 samples is varied to investigate the effect of Al number on MTO reactivity when compared to a SAPO-34 sample with only isolated Si Brønsted acid sites. These SSZ-13 samples display reduced transient selectivity behavior and extended reaction lifetimes as Si/Al increases; attributable to fewer paired Al sites. MTO reactivity for the higher Si/Al SSZ-13s resembles the SAPO-34 sample, suggesting that both catalysts owe their stable reaction behavior to isolated Brønsted acid sites.

Zeolites CHA and RHO are prepared without the use of organic structure-directing agents (OSDAs), dealuminated by steam treatments (500°C-800°C), and evaluated as catalysts for the MTO reaction. The effects of temperature and steam partial pressure during steaming are investigated. X-ray diffraction (XRD) and Ar physisorption show that steaming causes partial structural collapse of the zeolite, with degradation increasing with steaming temperature. ²⁷Al MAS NMR spectra of steamed materials reveal the presence of tetrahedral, pentacoordinate, and hexacoordinate aluminum.

Proton forms of as-synthesized CHA (Si/Al=2.4) and RHO (Si/Al=2.8) rapidly deactivate under MTO testing conditions (400°C, atmospheric pressure). CHA samples steamed at 600°C performed best among samples tested, showing increased olefin selectivities and catalyst lifetime. Acid washing these steamed samples further improved activity. Reaction results for RHO were similar to CHA, with the RHO sample steamed at 800°C producing the highest light olefin selectivities. Catalyst lifetime and C₂-C₃ olefin selectivities increase with increasing reaction temperature for both CHA-type and RHO-type steamed samples.

Dinâmica de uma floresta tropical manejada na Amazônia oriental

Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Florestal, 2016.

Ensemble Stream Model for Data-Cleaning in Sensor Networks

Ensemble Stream Modeling and Data-cleaning are sensor information processing systems have different training and testing methods by which their goals are cross-validated. This research examines a mechanism, which seeks to extract novel patterns by generating ensembles from data. The main goal of label-less stream processing is to process the sensed events to eliminate the noises that are uncorrelated, and choose the most likely model without over fitting thus obtaining higher model confidence. Higher quality streams can be realized by combining many short streams into an ensemble which has the desired quality. The framework for the investigation is an existing data mining tool. First, to accommodate feature extraction such as a bush or natural forest-fire event we make an assumption of the burnt area (BA*), sensed ground truth as our target variable obtained from logs. Even though this is an obvious model choice the results are disappointing. The reasons for this are two: One, the histogram of fire activity is highly skewed. Two, the measured sensor parameters are highly correlated. Since using non descriptive features does not yield good results, we resort to temporal features. By doing so we carefully eliminate the averaging effects; the resulting histogram is more satisfactory and conceptual knowledge is learned from sensor streams. Second is the process of feature induction by cross-validating attributes with single or multi-target variables to minimize training error. We use F-measure score, which combines precision and accuracy to determine the false alarm rate of fire events. The multi-target data-cleaning trees use information purity of the target leaf-nodes to learn higher order features. A sensitive variance measure such as f-test is performed during each node’s split to select the best attribute. Ensemble stream model approach proved to improve when using complicated features with a simpler tree classifier. The ensemble framework for data-cleaning and the enhancements to quantify quality of fitness (30% spatial, 10% temporal, and 90% mobility reduction) of sensor led to the formation of streams for sensor-enabled applications. Which further motivates the novelty of stream quality labeling and its importance in solving vast amounts of real-time mobile streams generated today.

Bland rutschbanebrallor och halkiga stockar : En studie om hur förskollärare tillsammans med barnen arbetar utomhus med fysikaliska begrepp som återkommer i läroplanen för åk 1-3.

The purpose of this study was to find out if preschool teachers use physical concepts included in thecurriculum for grade 1-3 (gravity, center of gravity, equilibrium, balance and friction) with the children in the outdoor environment at the preschool, if preschool teachers believe it is concepts that works to use with children in preschool outdoor environment and if the work could be developed regarding these physical concepts. To conduct the study, observations were done of preschool yards and semi-structured interviewswere conducted with preschool teachers who work in preschool right now. The result shows that it requires little resources to work with these concepts outdoors and teachers are positive to use the concepts but they do not use it to the extent they would like for a variety of reasons. In various ways, the preschool teachers also expresses that it is possible for children to absorb the physical concepts at preschool and it would help to later build a foundation in the form of childrens thinking that science is something fun and exciting that belong in their everyday lives.

Efeito dos processos de confeção nos produtos hortícolas

Trabalho Complementar apresentado à Universidade Fernando Pessoa como parte dos requisitos para obtenção do grau de licenciada em Ciências da Nutrição

Sensible use of primary energy in organic greenhouse production.

This document addresses the direct and indirect use of energy in European organic greenhouse horticulture (OGH) with the aim of reviewing available means for making it more environmental friendly and identifying knowledge gaps that should be addressed to attain this aim. The first observation is that there is no common regulation for energy use in OGH, which is not unexpected, since the need for climatisation is not uniformly distributed in the EU (and outside). Accordingly, the EU directive on organic agriculture does not set limitations on the use of energy, but rather promotes the responsible use of energy and of natural resources. The restrictions and rules of most private standards are slightly more stringent. Some standards have specific restrictions on the amount and sources of energy and/or on the seasonal use of energy for heating. Some standards also address processes that may affect (in)direct energy use, such as cultivation methods, mulching, lighting and growing media or substrates. However, most private standards have no or little restrictions or regulations on energy use. Accordingly, it should not surprise that very little quantitative information is available about energy use in OGH. In the present document we have filled the gaps with data with estimates drawn on energy use in conventional greenhouses. With respect to ongoing research, whereas many of the present research results about energy use and saving in conventional greenhouses are relevant (and also applied) in OGH, little research is devoted to address the energy use that is peculiar to OGH, particularly energy use for humidity control. In short, there are still a lot of knowledge gaps to improve quality and to lower energy use in organic greenhouses. The purpose of this document is a summary of present relevant knowledge about energy use and energy saving and of the perspective for improvement. In particular, the goal is to make an overview on the methods and technologies which can be used to reduce the energy use in OGH. We start from the assumption that methods and technologies that are used for reducing direct and indirect energy in conventional greenhouses can also be applied in organic greenhouses. Research on reducing energy use in conventional greenhouses is also more widely available because the area of conventional greenhouse horticulture is much larger than the area of OGH. When implementing these methods and techniques we should take into account the specific characteristics of organic agriculture like soil-based cultivation, use of organic fertilizers and the limited use of crop protection products. This document is organised as follows: first we report the results of a survey about energy use and relevant standards in the countries participating to the COST action (chapter 1); then we review the energy use for climatisation: heating (chapter 2) and humidity (chapter 3). In chapter 4 we review the available design and management means that would either reduce energy use and/or increase energy use efficiency by increasing productivity of OGH. In chapter 5 we present a short summary of existing information on indirect energy use, that is the energy required to manufacture production means (greenhouse structure and cover, fertilisers, equipment etc.) and for crop protection, particularly steaming, and briefly discuss possible savings. Finally (chapter 6) we review briefly the potential for application of renewable energy sources in OGH.

The genre personal diary: how the intimate is built

According to Bakhtin (1997), is as the genre entered into a speech, linked to certain field of human activity, that the subject of language and appropriates constitutes. Some of these fields have more close with the daily activities of individuals than others and produce more malleable, genres that allow greater individual interventions, as in the case of personal journal. Due to the intimate character of this genre, it attracted our attention and invited us to examine it more closely. In this paper, we investigate the personal diary genre, from the analysis of two documents written by ordinary people, in an attempt to learn a little bit more about the genre, of the writers and how their relationship with the language. To this end, two adults female people, provided us with their personal logs for analysis. Theoretically, we rely on the assumptions of Mikhail Bakhtin (1997) on the genres of discourse and the work developed by researcher Philippe Lejeune (2008) about autobiographical texts written by ordinaries people. The analysis of the diaries showed that, in addition to record experiences and feelings of the people, they were used as instrument of prayer to one of the person and as a means of creating a public image of himself, to the other.