1 |
Incomplete and Uncertain Information in Relational DatabasesZimanyi, Esteban 01 January 1992 (has links)
<p align="justify">In real life it is very often the case that the available knowledge is imperfect in the sense that it represents multiple possible states of the external world, yet it is unknown which state corresponds to the actual situation of the world. Imperfect knowledge can be of two different categories. Knowledge is incomplete if it represents different states, one of which is true in the external world. On the contrary, knowledge is uncertain if it represents different states which may be satisfied or are likely to be true in the external world.</p>
<p align="justify">Imperfect knowledge can be considered under two different perspectives: using either an algebraic or a logical approach. We present both approaches in relation with the standard relational model, providing the necessary background for the subsequent development.</p>
<p align="justify">The study of imperfect knowledge has been an active area of research, in particular in the context of relational databases. However, due to the complexity of manipulating imperfect knowledge, little practical results have been obtained so far. In this thesis we provide a survey of the field of incompleteness and uncertainty in relational databases;it can be used also as an introductory tutorial for understanding the intuitive semantics and the problems encountered when representing and manipulating such imperfect knowledge. The survey concentrates in giving an unifying presentation of the different approaches and results found in the literature, thus providing a state of the art in the field.</p>
<p align="justify">The rest of the thesis studies in detail the manipulation of one type of incomplete knowledge, namely disjunctive information, and one type of uncertain knowledge, namely probabilistic information. We study both types of imperfect knowledge using similar approaches, that is through an algebraic and a logical framework. The relational algebra operators are generalized for disjunctive and probabilistic relations, and we prove the correctness of these generalizations. In addition, disjunctive and probabilistic databases are formalized using appropriate logical theories and we give sound and complete query evaluation algorithms.</p>
<p align="justify">A major implication of these studies is the conviction that viewing incompleteness and uncertainty as different facets of the same problem would allow to achieve a deeper understanding of imperfect knowledge, which is absolutely necessary for building information systems capable of modeling complex real-life situations. </p>
|
2 |
Transparência na política monetária: teoria, empírico e projeçãoLima, Daniela Cunha de 02 October 2017 (has links)
Submitted by Daniela Lima (dclima@gmail.com) on 2017-10-23T00:56:49Z
No. of bitstreams: 1
Tese_DanielaLima.pdf: 1375366 bytes, checksum: be9b3161081f9ca9f80aed6d22ceda87 (MD5) / Approved for entry into archive by Suzinei Teles Garcia Garcia (suzinei.garcia@fgv.br) on 2017-10-23T16:40:57Z (GMT) No. of bitstreams: 1
Tese_DanielaLima.pdf: 1375366 bytes, checksum: be9b3161081f9ca9f80aed6d22ceda87 (MD5) / Made available in DSpace on 2017-10-24T11:42:30Z (GMT). No. of bitstreams: 1
Tese_DanielaLima.pdf: 1375366 bytes, checksum: be9b3161081f9ca9f80aed6d22ceda87 (MD5)
Previous issue date: 2017-10-02 / Over the last few years we have seen a huge evolution in the way central banks communicate around the World. Besides providing information to all, transparency allows agents to coordinate and is considered important in the control of inflation expectations. Despite the progress towards greater transparency, there are questions about the extent to which it is really desirable from the social point of view. This thesis analyzes the decision to increase transparency in an incomplete information environment, the empirical effect of transparency on inflation and assesses the predictive capacity of inflation models. The first chapter develops a model for an incomplete information environment. From a stylized game where coordination is desirable from an individual point of view, agents use public and private signals to choose their actions. The process of private information acquisition is endogenous and agents choose how much to invest in the accuracy of private information. In addition to the goal of stabilizing the economy, the monetary authority decides how much to reveal in public information. Public information is a two-edge instrument: it provides additional information and also coordinates expectation, serverving as a focal point for agent’s beliefs. However, by acting in high order belief, public information can over-coordinate agents, enhancing the damage of any errors. This chapter focuses on the following questions: Are there gains in increasing the accuracy of public information? In which situations does greater transparency increase social well-being and in what circumstances is opacity optimal? The results indicate that greater accuracy in public information can increase aggregate well-being, especially when information received privately by the government is not extremely accurate and when the cost of obtaining private information is an important channel. For cases where the information received by the government is extremely accurate, the opacity is optimal. The second chapter empirically analyzes the effects of central bank transparency on the level of inflation. From a panel of 100 countries, we assess whether greater transparency is associated with lower inflation, what types of transparency are most relevant, and especially if the effect of transparency on inflation is different between emerging and developed countries. The estimation faces the omitted variable problem: unobservable characteristics of a particular country can lead to both greater transparency and lower inflation. To try to control this endogeneity, we used three different methodologies: a two-step GMM (S-GMM) dynamic panel estimation developed by Arellano-Bover and Blundell-Bond, a fixed-effect panel and an ordinary least squares estimation. The results show that there is enough evidence that countries with greater transparency have lower inflation. Moreover, the effect of transparency in undeveloped countries is highly significant and negatively correlated with the level of inflation, while the effect for developed countries is smaller and less significant. Increased transparency in emerging countries may be related to regime changes, greater commitment to inflation control and credibility. Thus, greater transparency tends to have a significant impact on inflation in these emerging countries. When analyzing the five types of transparency that compose the index, the monetary policy transparency was the most frequently significant. This type of transparency is associated with timely explanations of monetary policy decisions and signals about the future interest trajectory, while the other sub-indices have more structural and bureaucratic characteristics with long-term effects. The third chapter compares the ability to project consumer inflation in Brazil (IPCA) outside the sample of three methodologies: the MIDAS, an augmented factor VAR (FAVAR) and a nowcast with mixed frequency model and dynamic factors. Over the last few years, several models of inflation projections have been suggested. In addition to the traditional time series models, new approaches allow the use of a large number of variables and the incorporation of samples at different frequencies in the same estimation without over-parameterization. In this chapter, we try to evaluate what kind of methodology is best to predict short-term inflation in Brazil. We estimated the models in four-year windows, with the out-of-sample forecast for the following year and forecast horizon one step ahead and compared the performance of the out-of-sample estimation done by the three models with a naive AR(1) model. The results show that the projections made using MIDAS are much more accurate than those estimated by FAVAR and nowcast, and all three methodologies were superior to AR(1). We noticed that there was a significant worsening of predictive capacity in the years 2015 and 2016 models, especially in the FAVAR and nowcast models. Also, we observed gains in accuracy in the combination of projections: the simple arithmetic mean of the projections has a smaller error than the individual projections performed in each estimation. / Ao longo dos últimos anos observamos uma enorme evolução na comunicação dos bancos centrais ao redor do mundo. Além do caráter informacional, a transparência atua na coordenação dos agentes e é considerada hoje importante aliada no controle de expectativas de inflação. A despeito do avanço no sentido de maior transparência, há questionamentos de até que ponto ela é realmente desejável do ponto de vista social. Essa tese analisa a decisão de aumentar a transparência em um ambiente de informação incompleta, o efeito empírico da transparência na inflação e avalia a capacidade preditiva de modelos de inflação. O primeiro capítulo desenvolve um modelo em ambiente de informação incompleta. A partir de um jogo estilizado em que a coordenação é desejável do ponto de vista individual, os agentes utilizam sinais públicos e privados para escolherem suas ações. O processo de aquisição de informação privada é endógeno e os agentes escolhem o quanto investir na precisão da informação privada a um custo linear. Além de ter por objetivo estabilizar a economia, a autoridade monetária decide o quanto revelar na informação pública. A informação pública tem um papel duplo: expandir o conjunto informacional e coordenar as expectativas dos agentes. Contudo, por atuar no high order belief, a informação pública pode coordenar excessivamente os agentes, potencializando os danos de eventuais erros. Esse capítulo foca nas seguintes questões: Há ganhos em aumentar a precisão da informação pública? Em que situações maior transparência aumenta o bem estar social e em quais circunstância a opacidade é ótima? Os resultados apontam que maior precisão na informação pública pode gerar aumento no bem estar agregado, especialmente quando a informação recebida privadamente pelo governo não é extremamente precisa e quando o custo de se obter informação privada é um canal importante. Para os casos em que a informação recebida pelo governo é extremamente precisa, a opacidade é ótimo. O segundo capítulo analisa empiricamente os efeitos da transparência do banco central no nível da inflação. A partir de um painel com 100 países, avaliamos se maior transparência está associada a menor inflação, quais tipos de transparência são mais relevantes e, principalmente, se o efeito da transparência na inflação é diferente entre países emergentes e desenvolvidos. A estimação enfrenta o problema de variável omitida: características não observáveis de determinado pais podem levar tanto à maior transparência, quanto à menor inflação. Para tentar controlar essa questão, utilizamos três metodologias diferentes: uma estimação de painel dinâmico via GMM (S-GMM) em dois passos desenvolvido por Arellano-Bover e Blundell-Bond, um painel de efeito fixo e uma estimação de mínimos quadrados ordinários. Os resultados apontam que há sim evidências de que países com maior transparência possuem menor inflação. Ainda, o efeito da transparência em países não desenvolvidos é altamente significante e negativamente correlacionado com nível da inflação, enquanto o efeito para países desenvolvidos é menor e pouco significante. Aumento de transparência em países emergentes pode estar relacionado a maior compromisso no controle inflacionário e credibilidade. Assim, maior transparência tende a ter impacto relevante na inflação desses países. Ao analisarmos os cinco tipos de transparência que compõe o índice, a transparência de política monetária foi a que se mostrou significante de forma mais frequente. Este tipo de transparência está associado a explicações tempestivas de decisões de política monetária e sinalizações sobre a trajetória de juros futura, enquanto os demais sub-índices possuem características mais estruturais e burocráticas, com efeitos de longo prazo. O terceiro capítulo compara a capacidade de projetar a inflação ao consumidor no Brasil (IPCA) fora da amostra de três metodologias: o MIDAS, um VAR aumentado com fatores (FAVAR) e um modelo de frequência mista e fatores dinâmicos de nowcast. Ao longo dos últimos anos, diversos modelos de projeções de inflação foram sugeridos. Em adição aos modelos tradicionais de séries de tempo, novas abordagens permitem o uso de um grande número de variáveis e a incorporação de amostras em diferentes frequências na mesma estimação sem que haja sobreparametrização. Neste capítulo, procuramos avaliar qual tipo de metodologia é melhor para prever a inflação de curto prazo no Brasil. Estimamos os modelos em janelas de quatro anos, com a previsão fora da amostra para o ano seguinte e horizonte de previsão um mês à frente e comparamos o desempenho dessa estimação fora da amostra dos três modelos com um modelo naive AR(1). Os resultados apontam que as projeções realizadas com a tecnologia MIDAS são muito mais acuradas do que àquelas estimadas pelo FAVAR e pelo nowcast, e todas as três metodologias se mostraram superiores ao AR(1). Percebemos que houve piora significativa na capacidade preditiva nos modelos anos de 2015 e 2016, principalmente nas projeções dos modelos FAVAR e nowcast. Ainda, observamos ganhos de acurácia na combinação de projeções: a média aritmética simples das projeções possui erro menor do que as projeções individuais realizadas em cada estimação.
|
3 |
Incomplete and uncertain information in relational databasesZimanyi, Esteban 01 January 1992 (has links)
<p align="justify">In real life it is very often the case that the available knowledge is imperfect in the sense that it represents multiple possible states of the external world, yet it is unknown which state corresponds to the actual situation of the world. Imperfect knowledge can be of two different categories. Knowledge is incomplete if it represents different states, one of which is true in the external world. On the contrary, knowledge is uncertain if it represents different states which may be satisfied or are likely to be true in the external world.</p><p><p align="justify">Imperfect knowledge can be considered under two different perspectives: using either an algebraic or a logical approach. We present both approaches in relation with the standard relational model, providing the necessary background for the subsequent development.</p><p><p align="justify">The study of imperfect knowledge has been an active area of research, in particular in the context of relational databases. However, due to the complexity of manipulating imperfect knowledge, little practical results have been obtained so far. In this thesis we provide a survey of the field of incompleteness and uncertainty in relational databases;it can be used also as an introductory tutorial for understanding the intuitive semantics and the problems encountered when representing and manipulating such imperfect knowledge. The survey concentrates in giving an unifying presentation of the different approaches and results found in the literature, thus providing a state of the art in the field.</p><p><p align="justify">The rest of the thesis studies in detail the manipulation of one type of incomplete knowledge, namely disjunctive information, and one type of uncertain knowledge, namely probabilistic information. We study both types of imperfect knowledge using similar approaches, that is through an algebraic and a logical framework. The relational algebra operators are generalized for disjunctive and probabilistic relations, and we prove the correctness of these generalizations. In addition, disjunctive and probabilistic databases are formalized using appropriate logical theories and we give sound and complete query evaluation algorithms.</p><p><p align="justify">A major implication of these studies is the conviction that viewing incompleteness and uncertainty as different facets of the same problem would allow to achieve a deeper understanding of imperfect knowledge, which is absolutely necessary for building information systems capable of modeling complex real-life situations. </p> / Doctorat en sciences, Spécialisation Informatique / info:eu-repo/semantics/nonPublished
|
Page generated in 0.1389 seconds