Search

Treball de fi de màster en Lingüística Teòrica i Aplicada. Director: Dr. Thomas Brochhagen Word embeddings represent word meaning in the form of a vector; however, the encoded information varies depending on the parameters the vector has been trained with. This paper analyzes how two parameters, context size and symmetry, influence word embedding information and aims to find if there exists a single distributional parametrization for capturing semantic similarity as well as relatedness. The models were trained with GloVe with different parametrizations; then, they were quantitatively evaluated through a similarity task, using WordSim-353 (for relatedness) and SimLex-999 (for semantic similarity) as benchmarks. The results show a minimal variation when manipulating some of the analyzed parameters, in particular between symmetric and asymmetric contexts, which leads us to conclude that it is not necessary to train models with large contexts for achieving good performance.

This dissertation focuses on the economic effects of hurricanes in the United States. Chapter 1 investigates the long-run consequences of prenatal hurricane exposure. Environmental conditions during gestation meaningfully affect human capital at birth, resulting in long-run economic consequences. We study the later-life effects of prenatal hurricane exposure for more than 1.4 million individuals born in the U.S. between 1980 and 1994, using administrative and survey data from the IRS and U.S. Census Bureau combined with satellite-derived hurricane data. We estimate that plausibly exogenous exposure to dangerous hurricane winds during gestation has no effects on later-life economic outcomes when hurricanes are forecasted, despite evidence from past work that hurricanes adversely affect birth outcomes. We also estimate that prenatal exposure to hurricanes outside of the forecasted area is associated with substantial losses: a $4,500 reduction in yearly adult earnings and a 5.4 percentile reduction in economic mobility. Back-of-the-envelope calculations suggest that in the absence of forecasts, total earnings lost from prenatal exposure would be $33 billion over the sample period. This highlights the value of advanced information in mitigating the translation of environmental shocks into economic damages. The second chapter, joint with Jonathan Colmer and John Voorheis, studies the distributional effects of hurricanes. We document that there has been no systematic movement towards or away from riskier areas over the last two decades and that non-Hispanic Black and less-educated individuals are significantly more likely to be exposed to hurricanes than other Americans. Second, we estimate the dynamic causal effects of hurricane exposure, finding that individuals exposed to hurricane winds experience modest reductions in total income and earnings in the immediate aftermath of a hurricane, with a return to trend within five years. Earnings losses are largely offset through increases in transfer payments. However, our results suggest that Black and Hispanic Americans are disproportionately affected by hurricane exposure, particularly low and high income individuals. We estimate that these individuals experience reductions in total income and earnings that are six times larger than the average effects. The third chapter studies the role of past experience with hurricanes. The relationship between hurricane exposure and economic damages will depend on decisions made, and these decisions cannot be perfectly optimized without perfect information. I study the role of hurricane experience by combining 37 years of satellite-derived hurricane exposure data with county-level income and defensive investments. I estimate that hurricane exposure in inexperienced counties reduces total income by about 1%, but past experience with hurricanes reduces this effect by about 30% per event. In addition, marginal defensive investments increase by about 18% per event experienced. These results may suggest that hurricane experience prompts learning and that people initially underinvest in self-protection against hurricanes.

L'intelligenza artificiale cerca di spiegare come gli agenti intelligenti si comportano. Il linguaggio è uno dei media di comunicazioni più importanti e studiare delle teorie che permettano di definire il significato di espressioni naturali è molto importante. I linguisti hanno usato con successo linguaggi artificiali basati su logiche, ma una theory che ha avuto un impatto significativo in intelligenza artificiale è la semantica distribuzionale. La semantica distribuzionale afferma che il significato di espressioni in linguaggio naturale può essere derivato dal contesto in cui tali espressioni compaiono. Questa teoria è stata implementata da algoritmi che permettono di generare rappresentazioni vettoriali delle espressioni del linguaggio natural in modo che espressioni simili vengano rappresentate con vettori simili. Negli ultimi anni, gli scienziati cognitivi hanno sottolineato che queste rappresentazioni sono correlate con l'associative learning e che sono anche in grado di catturare bias e stereotype del testo. Diventa quindi importante trovare metodologie per comparare rappresentazioni che arrivano da sorgenti diverse. Ad esempio, usare questi algoritmi su testi di periodi differenti genera rappresentazioni differenti: visto che il linguaggio muta nel tempo, trovare delle metododoloie per comparare come le parole si sono mosse è un task imporante per l'intelligenza artificiale (e.g., la parola "amazon" ha cambiato il suo significato principale negli ultimi anni) In questa tesi, introduciamo un modello comparative basato su testi che permette di comparare rappresentazioni di sorgenti diverse generate con la semantica distribuzionale. Proponiamo un modello che è efficiente ed efficace e mostriamo che possiamo anche gestire nomi di entità e non solo paorle, superando problemi legati all'ambiguità del linguaggio. Alla fine, mostriamo che è possibile combinare questi metodi con approcci logici e fare comparazioni utilizzando costrutti logici. One of the main goals of artificial intelligence is understanding how intelligent agent acts. Language is one of the most important media of communication, and studying theories that can account for the meaning of natural language expressions is an important task. Language is one of the most important media of communication, and studying theories that can account for the meaning of natural language expressions is a crucial task in artificial intelligence. Distributional semantics states that the meaning of natural language expressions can be derived from the context in which the expressions appear. This theory has been implemented by algorithms that generate vector representations of natural language expressions that represent similar natural language expressions with similar vectors. In the last years, several cognitive scientists have shown that these representations are correlated with associative learning and they capture cognitive biases and stereotypes as they are encoded in text corpora. If language is encoding important aspects of cognition and our associative knowledge, and language usage change across the contexts, the comparison of language usage in different contexts may reveal important associative knowledge patterns. Thus, if we want to reveal these patterns, we need ways to compare distributional representations that are generated from different text corpora. For example, using these algorithms on textual documents from different periods will generate different representations: since language evolves during time, finding a way to compare words that have shifted over time is a valuable task for artificial intelligence (e.g., the word "Amazon" has changed its prevalent meaning during the last years). In this thesis, we introduce a corpus-based comparative model that allows us to compare representations of different sources generated under the distributional semantic theory. We propose a model that is both effective and efficient, and we show that it can also deal with entity names and not just words, overcoming some problems that follow from the ambiguity of natural language. Eventually, we combine these methods with logical approaches. We show that we can do logical reasoning on these representations and make comparisons based on logical constructs.

Swiss targets for climate policy require significant reductions of emissions by 2050. While such reductions can be achieved in a cost-efficient manner by employing taxes on greenhouse gas emissions, such taxes tend to lead to a regressive distribution of policy cost among households. To counteract such a regressive outcome, tax revenue may be recycled in a progressive way. This paper uses a computable general equilibrium model coupled with a microsimulation of household income and expenditure to examine the policy cost of different carbon tax policies and their distribution across households. I find that in the absence of revenue recycling, emission taxation leads to a regressive distribution of policy cost. I analyze different revenue recycling schemes (per-capita lump-sum transfers, reductions in labor taxation, and reductions in VAT taxation of necessary commodities) and their ability to avoid regressive outcomes., Swiss Journal of Economics and Statistics, 155 (1), ISSN:0303-9692, ISSN:2235-6282

One of the defining traits of language is its capacity to mediate between concepts in our mind, which encapsulate generalizations, and the things they refer to in a given communicative act, with all their idiosyncratic properties. This article examines precisely this interplay between conceptual and referential aspects of meaning, and proposes that concept composition (or concept combination, a term more commonly used in Psychology) exploits both: Conceptually afforded composition is at play when a modifier and its head fit as could be expected given the properties of the two concepts involved, whereas in referentially afforded composition the result of the composition depends on specific, independently available properties of the referent. For instance, red box tends to be applied to boxes whose surface is red, but, given the appropriate context, it can also be applied to e.g. a brown box that contains red objects. We support our proposal with data from nominal modification, and explore a way to formally distinguish the two kinds of composition and integrate them into a more general framework for semantic analysis. Along the way, we recover the classically Fregean notion of sense as including conceptual information, and show the potential of distributional semantics, a framework that has become very influential in Cognitive Science and Computational Linguistics, to address research questions from a theoretical linguistic perspective. This project has received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement No 655577 (LOVe) and from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 715154, AMORE); Spanish MINECO grants FFI2010-15006, FFI2010-09464-E, and FFI2013-41301, as well as an ICREA Academia award to the first author.

This study assessed the likely economic, distributional, and fisheries resource impacts of a potential free trade agreement (FTA) between the Philippines and European Union (EU) on the fisheries sector of the former. The study used secondary data from institutional sources and results and findings of past studies. Among others, the study found that a) the elimination of tariffs will likely increase fisheries outputs and exports as well as help reduce poverty in the fisheries sector and the general population; b) the elimination of tariffs will likely diversify the currently limited country destinations and number of exported fisheries products of the Philippines to the EU; c) other than tariffs, there are nontariff measures (NTMs) that significantly impede freer flow of fisheries products from the Philippines to the EU that need to be considered; d) some participants in the Philippine fisheries sector will gain from an FTA while others will lose but the net benefits to the sector and economy is not known; and e) increase in fisheries exports due to the FTA will likely worsen fisheries resource overexploitation although the inflow of cheaper imported fish will tend to reduce the overexploitation.The study concludes, among others, that if a Philippines-EU FTA materializes it should not only reduce or eliminate tariffs in fisheries products but also the NTMs. It also argued that the government should provide safety nets for the fisheries participants who are going to be disadvantaged by the FTA and implement the needed resource and environmental management that will allow sustainable exploitation of fisheries resources even with increased trade.

The increasing use of energy from renewable sources (RE) for the generation of heat and electricity in Germany has also led to an increasingly intensive debate on its advantages and disadvantages. The discussion mostly centers on the cost effects, because beneficial effects often are harder to quantify. Benefits comprise indirect effects or effects which lie far in the future. Scientific studies also frequently focus on single aspects, which cannot be aggregated easily, because they occur in different sectors and comprise technology system-wide effects, distributional effects or overall macro-economic effects. This study intends to answer questions refering to clear definitions of effects when increasing renewable energy, methods of adding and balancing the effects, the choice of methodological approach, support mechanisms, suitable time span or spatial system border. After a brief sketch of the basic methods applied, the contribution will present estimates for a wide range of effects. It closes with suggestions on possibilities how to add and balance the effects.

Este trabajo aplica un análisis de incidencia distributiva para estudiar los patrones de acceso y gasto en servicios básicos (educación, salud, transporte público, agua, electricidad, gas y telecomunicaciones) en los países Latinoamericanos. Se encuentra que los gastos de los hogares en estos servicios presentan un patrón distributivo pro-rico, principalmente debido a que los hogares más pobres sufren limitaciones al acceso. Por otra parte, los servicios con mayor participación en el presupuesto familiar (educación, salud y transporte) se caracterizan por índices de Kakwani moderados a pequeños, mientras que los servicios con altos valores del índice de Kakwani (telecomunicaciones y gas) tienen una baja participación en el presupuesto, lo que sugiere que el efecto distributivo de potenciales reformas en estos sectores serían pequeños., Otra versión de este trabajo fue publicada en los Anales de la XLIV Reunión Anual de la Asociación Argentina de Economía Política, Mendoza, Argentina. ISSN 1852-0022, ISBN 978-987-99570-7-3., Centro de Estudios Distributivos, Laborales y Sociales (CEDLAS)

This thesis presents a new kind of model for meaning construction, within the framework of continuous mathematics. We take the language as a morphodynamic system following the basic principles of the Gestalttheorie. Linguistics units find their meaning in a semantic space with a differentiable variety structure. We assess the validity of this model by implementing and using it on a french adjectival lexicon. Our method of automatic building of the semantic spaces, which relies on the analysis of a graph of synonymy, enables us to explore the adjectival lexicon as a whole, or to construct local semantic spaces. Local semantic spaces are used as a basis of a dynamic method for meaning computation, which allows us to take the different factors of adjectival polysemy into account. The use of global semantic spaces is very promising, not only in the domain of meaning computation, but also in that of small world graphs exploration., Ce travail de thèse présente un modèle de construction du sens d'un genre nouveau, défini dans le cadre des mathématiques du continu. Le langage y est vu comme un système morphodynamique, obéissant aux principes de base de la Gestalttheorie. Les unités linguistiques découpent leur sens dans un espace sémantique possédant une structure de variété différentiable. Nous avons implémenté ce modèle et l'avons testé sur le lexique adjectival français. Une méthode de construction automatique des espaces sémantiques, reposant sur l'analyse d'un graphe de synonymie, permet d'explorer le lexique adjectival dans son ensemble, ou de construire des espaces locaux. Les espaces sémantiques locaux servent de base à une méthode dynamique de calcul du sens, permettant de prendre en compte les différents facteurs de polysémie adjectivale. L'utilisation des espaces sémantiques globaux ouvre de belles perspectives, tant dans le domaine du calcul du sens que celui de l'exploration de graphes petit monde.

Aircraft noise exposure is a health risk and there is evidence that noise annoyance partly mediates the association between noise exposure and stress-related health risks. Thus, approaches to reduce annoyance may be beneficial for health. Annoyance is influenced by manifold non-acoustic factors and perceiving a fair and trustful relationship between the airport and its residents may be one of them. The distribution of aircraft noise exposure can be regarded as a fairness dilemma: while residents living near an airport may seem to have some advantages, the majority of residents living under certain flight routes or in their immediate proximity suffer from the disadvantages of the airport, especially the noise. Moreover, a dilemma exists between the airport’s beneficial economic impact for a region and the physical and psychological integrity of residents. Aircraft noise exposure through the lens of social justice research can help to improve our understanding of noise annoyance. Research indicates that the fairness perceptions of the parties involved can be enhanced by (a) improving individual cost–benefit ratios, (b) providing a fair procedure for deciding upon the noise distribution, and (c) implementing fair social interaction with residents. Based on the review of evidence from social justice research, we derive recommendations on how fairness aspects can be integrated into aircraft noise management with the purpose of improving the relationship between the airport and its residents, to reduce annoyance, and to enhance the acceptance of local aviation and the airport as a neighbor.