Hasil untuk "Latin America. Spanish America"

Brigitte A. Mora-Reyes, Jennifer A. Drewyor, Abel A. Reyes-Angulo

Artificial intelligence (AI) systems often reflect biases from economically advanced regions, marginalizing contexts in economically developing regions like Latin America due to imbalanced datasets. This paper examines AI representations of diverse Latin American contexts, revealing disparities between data from economically advanced and developing regions. We highlight how the dominance of English over Spanish, Portuguese, and indigenous languages such as Quechua and Nahuatl perpetuates biases, framing Latin American perspectives through a Western lens. To address this, we introduce a culturally aware dataset rooted in Latin American history and socio-political contexts, challenging Eurocentric models. We evaluate six language models on questions testing cultural context awareness, using a novel Cultural Expressiveness metric, statistical tests, and linguistic analyses. Our findings show that some models better capture Latin American perspectives, while others exhibit significant sentiment misalignment (p < 0.001). Fine-tuning Mistral-7B with our dataset improves its cultural expressiveness by 42.9%, advancing equitable AI development. We advocate for equitable AI by prioritizing datasets that reflect Latin American history, indigenous knowledge, and diverse languages, while emphasizing community-centered approaches to amplify marginalized voices.

Guido Ivetta, Pietro Palombini, Sofía Martinelli et al.

The evaluation of societal biases in NLP models is critically hindered by a geo-cultural gap, This leaves regions such as Latin America severely underserved, making it impossible to adequately assess or mitigate the perpetuation of harmful regional stereotypes in language technologies. This paper presents LACES, a stereotype association dataset, for 15 Latin American countries. This dataset includes 4,789 stereotype associations manually created and annotated by 83 participants. The dataset was developed through targeted community partnerships across Latin America. Additionally, in this paper, we propose a novel adaptive data collection methodology that uniquely integrates the sourcing of new stereotype entries and the validation of existing data within a single, unified workflow. This approach results in a resource with more unique stereotypes than previous static collection methods, enabling a more efficient stereotype collection. The paper further supports the quality of LACES by demonstrating reduced efficacy of debiasing methods on this dataset in comparison to existing popular stereotype benchmarks.

Tim Whittaker, Alejandro Di Luca, Francois Roberge et al.

We evaluate the performance of various configurations of the Canadian Regional Climate Model (CRCM6-GEM5) in simulating 10-meter wind speeds using data from 27 AmeriFlux stations across North America. The assessment employs a hierarchy of error metrics, ranging from simple mean bias to advanced metrics that account for the dependence of wind speeds on variables such as friction velocity and stability. The results reveal that (i) the value of roughness length (z0) has a large effect on the simulation of wind speeds, (ii) using a lower limit for the Obhukov length instead of a lower limit for the lowest level wind speed seems to deteriorate the simulation of wind speeds under very stable conditions, (iii) the choice of stability function has a small but noticeable impact on the wind speeds, (iv) the turbulent orographic form drag scheme shows improvement over effective roughness length approach.

Andrea Busto-Castiñeira, Francisco J. González-Castaño, Silvia García-Méndez et al.

In recent years, the field of Natural Language Generation (NLG) has been boosted by the recent advances in deep learning technologies. Nonetheless, these new data-intensive methods introduce language-dependent disparities in NLG as the main training data sets are in English. Also, most neural NLG systems use decoder-only (causal) transformer language models, which work well for English, but were not designed with other languages in mind. In this work we depart from the hypothesis that they may introduce generation bias in target languages with less rigid word ordering, subject omission, or different attachment preferences for relative clauses, so that for these target languages other language generation strategies may be more desirable. This paper first compares causal and non-causal language modeling for English and Spanish, two languages with different grammatical structures and over 1.5 billion and 0.5 billion speakers, respectively. For this purpose, we define a novel metric of average causal and non-causal context-conditioned entropy of the grammatical category distribution for both languages as an information-theoretic a priori approach. The evaluation of natural text sources (such as training data) in both languages reveals lower average non-causal conditional entropy in Spanish and lower causal conditional entropy in English. According to this experiment, Spanish is more predictable than English given a non-causal context. Then, by applying a conditional relative entropy metric to text generation experiments, we obtain as insights that the best performance is respectively achieved with causal NLG in English, and with non-causal NLG in Spanish. These insights support further research in NLG in Spanish using bidirectional transformer language models.

Carolina Rodríguez Tsouroukdissian

Abstract. This article examines early translations of “A cartomante,” one of the most anthologized stories written by Brazilian author Joaquim Maria Machado de Assis. I compare an Argentine translation and an English translation vis-à-vis the Portuguese original to determine to what extent they preserve and reproduce the literary features of the original text. I assess the alterations of the authorial voice in terms of additions, omissions, word choice, and style. Translation studies notions developed by Ernst-August Gutt and Lin Zhu inform this analysis. Whereas the Argentine translation tends to present more typos, suppress words, and reduce the intensity of some passages, the English translation over-dramatizes and over-explains some scenes, at the same time that it reimagines the characters to make them more attractive to the US readership. However, both translations reject some of the most characteristic aspects of Machado de Assis’s writing such as colloquialism and self-reflexivity. The close reading of these translations can help improve our understanding of Machado de Assis’s reception in Latin America and the United States at the beginning of the 20th century.

Divyansh Chug, Francina Dominguez, Christopher M Taylor et al.

Soil moisture-precipitation (SM-PPT) feedbacks at the mesoscale represent a major challenge for numerical weather prediction, especially for subtropical regions that exhibit large variability in surface SM. How does surface heterogeneity, specifically mesoscale gradients in SM and land surface temperature (LST), affect convective initiation (CI) over South America? Using satellite data, we track nascent, daytime convective clouds and quantify the underlying antecedent (morning) surface heterogeneity. We find that convection initiates preferentially on the dry side of strong SM/LST boundaries with spatial scales of tens of kilometers. The strongest alongwind gradients in LST anomalies at 30 km length scale underlying the CI location occur during weak background low-level wind (<2.5m/s), high convective available potential energy (>1500J/kg) and low convective inhibition (<250J/kg) over sparse vegetation. At 100 km scale, strong gradients occur at the CI location during convectively unfavorable conditions and strong background flow. The location of PPT is strongly sensitive to the strength of the background flow. The wind profile during weak background flow inhibits propagation of convection away from the dry regions leading to negative SM-PPT feedback whereas strong background flow is related to longer lifecycle and rainfall hundreds of kilometers away from the CI location. Thus, the sign of the SM-PPT feedback is dependent on the background flow. This work presents the first observational evidence that CI over subtropical South America is associated with dry soil patches on the order of tens of kilometers. Convection-permitting numerical weather prediction models need to be examined for accurately capturing the effect of SM heterogeneity in initiating convection over such semi-arid regions.

Maeve Upton, Andrew Parnell, Andrew Kemp et al.

We propose a Bayesian, noisy-input, spatial-temporal generalised additive model to examine regional relative sea-level (RSL) changes over time. The model provides probabilistic estimates of component drivers of regional RSL change via the combination of a univariate spline capturing a common regional signal over time, random slopes and intercepts capturing site-specific (local), long-term linear trends and a spatial-temporal spline capturing residual, non-linear, local variations. Proxy and instrumental records of RSL and corresponding measurement errors inform the model and a noisy-input method accounts for proxy temporal uncertainties. Results focus on the decomposition of RSL over the past 3000 years along the Atlantic coast of North America.

Gabriel Tiberiu Pană, Alexandru Nicolin-Żaczek

We present a detailed description of seismic activity in Romania, Italy, and Japan, as well as the California seismic zone in the United States of America, based on the statistical analysis of the underlying earthquake networks used to model the aforementioned zones. Our results on network connectivity and simple network motifs allow for a complex description of seismic zones, while at the same time reinforcing the current understanding of seismicity as a critical phenomenon. The reported distributions on node connectivity, three-, and four-event motifs are consistent with power-law, i.e., scale-free, distributions over large intervals and are robust across earthquake networks obtained from different discretizations of the seismic zones of interest. In our analysis of the distributions of node connectivity and simple motifs, we distinguish between the global distribution and the powerlaw part of it with the help of maximum likelihood estimation (MLE) method and complementary cumulative distribution functions (CCDF). The main message is that the distributions reported for the aforementioned seismic zones have large power-law components, extending over some orders of magnitude, independent of discretization. All the results were obtained using publicly-available databases and open-source software, as well as a new toolbox available on GitHub, specifically designed to automatically analyze earthquake databases.

Marianne Palisse

Ferreira, Paula Botafogo Caricchio

A partir da trajetória de João Soares Lisboa, único dos réus “condenado por conluio republicano” na primeira devassa política do Império do Brasil, o artigo trata da construção e derrota política de projetos de Brasil que rivalizavam com o da centralização no governo do Rio de Janeiro, sob a dinastia de Bragança. Depois de condenado ao exílio do Brasil, Soares Lisboa aportou em Recife e se juntou à Confederação do Equador (1824), publicando o Desengano aos Brasileiros para “desenganar” seus leitores sobre o protagonismo da dinastia Bragança na história da independência do Brasil. Ao refletir sobre os eventos da história da independência do Brasil e a construção do projeto monárquico do Império pela dinastia de Bragança, Soares Lisboa disseminou no Desengano os princípios de outros projetos de Estado e de nação do Brasil que haviam sido derrotados até então e que ele defendia na organização do “governo Monarco-Democrático” no Correio do Rio de Janeiro em 1822-1823 e que em 1824 só via possível em uma república, tal como propunham os ideólogos da Confederação do Equador (1824). No Desengano, contestava o constitucionalismo de D. Pedro I e de D. João VI, descrevendo uma espécie de natureza dos monarcas que era incompatível com um governo de liberdade. Essas publicações e a participação na Confederação do Equador impactaram a memória sobre a trajetória de João Soares Lisboa na literatura histórica e esta funcionou como referencial para qualificar a radicalidade de outros personagens do período

Caroline S. Wagner, Xiaojing Cai

The COVID-19 global pandemic starting in January 2020 disrupted international collaborations in scholarly exchange, reducing mobility and connections across the globe. An examination of Web of Science-indexed publications from China, the European Union-28 and the United States of America shows a drop in publications numbers coming from the EU-28 and the United States in 2021. Importantly, cooperation between China and the United States drops without a corresponding drop between China and the EU-28. Moreover, the drop in China-USA cooperation can be seen beginning in 2019, before the pandemic, at a time when political tensions around science, technology, and innovation arose, with the United States claiming that China was violating intellectual property norms. The patterns suggest that political tensions, more than the pandemic, influenced the drop in China-USA cooperation.

Javier Huertas-Tato, Alejandro Martin, David Camacho

The appearance of complex attention-based language models such as BERT, Roberta or GPT-3 has allowed to address highly complex tasks in a plethora of scenarios. However, when applied to specific domains, these models encounter considerable difficulties. This is the case of Social Networks such as Twitter, an ever-changing stream of information written with informal and complex language, where each message requires careful evaluation to be understood even by humans given the important role that context plays. Addressing tasks in this domain through Natural Language Processing involves severe challenges. When powerful state-of-the-art multilingual language models are applied to this scenario, language specific nuances use to get lost in translation. To face these challenges we present \textbf{BERTuit}, the larger transformer proposed so far for Spanish language, pre-trained on a massive dataset of 230M Spanish tweets using RoBERTa optimization. Our motivation is to provide a powerful resource to better understand Spanish Twitter and to be used on applications focused on this social network, with special emphasis on solutions devoted to tackle the spreading of misinformation in this platform. BERTuit is evaluated on several tasks and compared against M-BERT, XLM-RoBERTa and XLM-T, very competitive multilingual transformers. The utility of our approach is shown with applications, in this case: a zero-shot methodology to visualize groups of hoaxes and profiling authors spreading disinformation. Misinformation spreads wildly on platforms such as Twitter in languages other than English, meaning performance of transformers may suffer when transferred outside English speaking communities.

Martín L. E. Wasserman

A lo largo del último tercio del siglo XVIII el crecimiento económico y la incertidumbre se conjugaban en Buenos Aires, cuya interacción comercial continuaba apoyándose en distintas prácticas de endeudamiento, tal como sucedía desde la emergencia de la ciudad. Para afrontar los riesgos inherentes a la interacción crediticia, los actores podían depositar su crédito en tomadores confiables, cuya correspondencia se presumía asegurada por la proximidad relacional entre las partes. Pero el caso de Buenos Aires también ha dado cuenta de la existencia de transacciones que no se circunscribían a acreedores y deudores pertenecientes a un mismo círculo de sociabilidad: el crédito podía depositarse, asimismo, en un mediador cuya fiabilidad atenuase el riesgo de la distancia relacional entre acreedor y deudor. Y esa mediación habría de apoyarse, por lo tanto, en la capacidad del intermediario para gestionar eficientemente la información crediticia sobre el deudor y, con ello, los riesgos de la operación. El presente artículo tiene por objetivo indagar el rol que los escribanos porteños tuvieron como intermediarios financieros en el mercado de créditos durante el último tercio del siglo XVIII. Se realiza un análisis estadístico sobre los abecedarios de cada protocolo notarial y se evalúa la importancia de la información en su desempeño.

Rui Wang, Jiahui Chen, Guo-Wei Wei

The importance of understanding SARS-CoV-2 evolution cannot be overemphasized. Recent studies confirm that natural selection is the dominating mechanism of SARS-CoV-2 evolution, which favors mutations that strengthen viral infectivity. We demonstrate that vaccine-breakthrough or antibody-resistant mutations provide a new mechanism of viral evolution. Specifically, vaccine-resistant mutation Y449S in the spike (S) protein receptor-bonding domain (RBD), which occurred in co-mutation [Y449S, N501Y], has reduced infectivity compared to the original SARS-CoV-2 but can disrupt existing antibodies that neutralize the virus. By tracing the evolutionary trajectories of vaccine-resistant mutations in over 1.9 million SARS-CoV-2 genomes, we reveal that the occurrence and frequency of vaccine-resistant mutations correlate strongly with the vaccination rates in Europe and America. We anticipate that as a complementary transmission pathway, vaccine-resistant mutations will become a dominating mechanism of SARS-CoV-2 evolution when most of the world's population is vaccinated. Our study sheds light on SARS-CoV-2 evolution and transmission and enables the design of the next-generation mutation-proof vaccines and antibody drugs.

Julian Posada

This presentation for the AIES 21 doctoral consortium examines the Latin American crowdsourcing market through a decolonial lens. This research is based on the analysis of the web traffic of ninety-three platforms, interviews with Venezuelan data workers of four platforms, and the analysis of the documentation issued by these organizations. The findings show that (1) centuries-old global divisions of labor persist, in this case, with requesters located in advanced economies and workers in the Global South. (2) That the platforms' configuration of the labor process constrains the agency of these workers when producing annotations. And, (3) that ideologies originating from the Global North serve to legitimize and reinforce this global labor market configuration.

Aline Miranda

Este texto foi proferido na Conferência de abertura do III Negras Antropologias, realizado nos dias 26 e 27 de novembro de 2019, no Instituto de Ciências Sociais da Universidade de Brasília. Primeiro apresento um breve histórico sobre a formação do Coletivo Zora Hurston, composto por estudantes negras e negros do Programa de Pós-Graduação em Antropologia Social da UnB, e relembro as edições anteriores do evento. Em seguida, ancorada nos trabalhos de Frantz Fanon, Abdias do Nascimento e Lélia Gonzalez, discuto os efeitos de suas negras epistemologias na teoria antropológica.

Josimar Chire-Saire, Khalid Mahmood

As of July 31, 2020, the COVID-19 pandemic has over 17 million reported cases, causing more than 667,000 deaths. Countries irrespective of economic status have succumbed to this pandemic. Many aspects of the lives, including health, economy, freedom of movement have been negatively affected by the coronavirus outbreak. Numerous strategies have been taken in order to prevent the outbreak. Some countries took severe resections in the form of full-scale lockdown, while others took a moderate approach of dealing with the pandemics, for example, mass testing, prohibiting large-scale public gatherings, restricting international travels. South America adopted primarily the lockdown strategies due to inadequate economy and health care support. Since the social interactions between the people are primarily affected by the lockdown, psychological distress, e.g. anxiety, stress, fear are supposedly affecting the South American population in a severe way. This paper aims to explore the impact of lockdown over the psychological aspect of the people of all the Spanish speaking South American capitals. We have utilized infodemiology approach by employing large-scale Twitter data-set over 33 million feeds in order to understand people's interaction over the months of this on-going coronavirus pandemic. Our result is surprising: at the beginning of the pandemic, people demonstrated strong emotions (i.e. anxiety, worry, fear) which declined over time even though the actual pandemic is worsening by having more positive cases, and inflicting more deaths. This leads us to speculate that the South American population is adapting to this pandemic thus improving the overall psychological distress.

Hyeju Jang, Emily Rempel, Giuseppe Carenini et al.

Social media is a rich source where we can learn about people's reactions to social issues. As COVID-19 has significantly impacted on people's lives, it is essential to capture how people react to public health interventions and understand their concerns. In this paper, we aim to investigate people's reactions and concerns about COVID-19 in North America, especially focusing on Canada. We analyze COVID-19 related tweets using topic modeling and aspect-based sentiment analysis, and interpret the results with public health experts. We compare timeline of topics discussed with timing of implementation of public health interventions for COVID-19. We also examine people's sentiment about COVID-19 related issues. We discuss how the results can be helpful for public health agencies when designing a policy for new interventions. Our work shows how Natural Language Processing (NLP) techniques could be applied to public health questions with domain expert involvement.

Ana Gallego-Cuiñas, Esteban Romero-Frías, Wenceslao Arroyo-Machado

The present paper uses Twitter to analyze the current state of the worldwide, Spanish-language, independent publishing market. The main purposes are to determine whether certain Latin American Spanish-language independent publishers function as gatekeepers of World Literature and to analyze the geopolitical structure of this global market, addressing both the Europe-America dialectic and neocolonial practices. After selecting the sample of publishers, we conducted a search for their Twitter profiles and located 131; we then downloaded data from the corresponding Twitter APIs. Finally, we applied social network analysis to study the presence of and interaction between our sample of independent publishers on this social media. Our results provide data-based evidence supporting the hypothesis of some literary critics who suggest that in Latin America, certain publishers act as gatekeepers to the mainstream book market. Therefore, Twitter could be considered a valid source of information to address the independent book market in Spanish. By extension, this approach could be applied to other cultural industries in which small and medium-sized agents develop a digital presence in social media. This paper combines social network analysis and literary criticism to provide new evidence about the Spanish-language book market. It helps validate the aforementioned hypothesis, proposed by literary critics, and opens up new paths along which to pursue an interpretative, comparative analysis.

Jessica Byron

The article explores the implications of BREXIT for CARIFORUM states and UK Caribbean territories. It stresses that all conclusions are subject to change since the BREXIT talks are ongoing and the situation is fluid. It shows the diverse levels of political or economic vulnerability to BREXIT experienced by different territories and examines the options for future political and economic relations with the United Kingdom and the European Union.