62880 research outputs found
Sort by
Happiness prediction with domain knowledge integration and explanation consistency
No full textHappiness prediction based on large-scale online data and machine learning models is an emerging research topic that underpins a range of issues, from personal growth to social stability. Many advanced machine learning (ML) models with explanations are used for happiness online assessment while maintaining high accuracy of results. However, expert feedback and sociological theory may be absent from these models, which limits the association between prediction results and the right reasons for why they occurred. Sociological studies have shown that primary and secondary relations are inherent in happiness factors, which can be used as domain knowledge to guide model training. Inspired by such insights, this article attempts to provide new insights into the explanation consistency from an empirical study perspective. Then this article studies how to represent and introduce domain knowledge constraints to make ML models more trustworthy. We achieve this by 1) proving that multiple prediction models with additive factor attributions will have the desirable property of primary and secondary relations consistency; and 2) showing that factor relations with quantity can be represented as an importance distribution for encoding domain knowledge. Factor explanation difference is penalized by the Wasserstein distance among prediction models. Experimental results using two online datasets show that domain knowledge of stable factor relations exists. Using this knowledge not only improves happiness prediction accuracy but also reveals more significant happiness factors for assisting decisions
AI reshaping financial modeling
Get PDFThis perspective paper argues that, given the strong theoretical foundations and clear economic interpretations of traditional financial models, the integration of artificial intelligence (AI) in finance should prioritize enhancing these models—by incorporating alternative data sources and recalibrating key variables—rather than replacing them with opaque, albeit accurate, black-box models. We summarize studies that follow this approach, with a focus on the cases of Capital Asset Pricing Model (CAPM), Markowitz Mean-Variance Optimization (MVO), and the Black-Litterman Model (BLM). We demonstrate how AI, particularly Natural Language Processing (NLP) models, enables dynamic input estimation, nonlinear pattern discovery, sentiment extraction from financial text and sentiment-aware forecasting, and improved risk modeling, thereby addressing longstanding limitations in traditional frameworks. In addition, we highlight how this approach to some extent preserves interpretability—essential for regulatory compliance and investor trust—by tracing model decisions to intuitive, often human-understandable sources of information. By augmenting rather than replacing financial theory, this approach not only improves empirical performance but also enriches theoretical understanding, marking a paradigm shift in how financial models are built, explained, and applied
Ocean turbulent heat flux responses to sea surface salinity variability during Benguela Niños and Niñas
Get PDFBenguela Niño and Niña events are episodes of extreme warming and cooling off Angola with impacts on fisheries, ecosystems, and rainfall in southwest Africa. They are typically forced remotely or locally by variations in equatorial or alongshore winds, respectively. We use an extensive in‐situ data set to show that sea surface salinity (SSS) changes can also act as a local forcing that amplifies these extreme warm and cold events by altering the water column stratification and consequently the impact of subsurface mixing. The mixed layer turbulent heat loss during an extreme warm episode with unusually low SSS in 1995 is nearly 3× lower than during a cold event with high SSS in 1997. We also demonstrate that interannual turbulent heat flux variability in early boreal spring off Angola is strongly impacted by salt advection fluctuations, and that this turbulent mixing is significant for altering mixed layer temperatures and restoring its salinities
The first direct evidence of the Djeitun pottery type in Eastern Mazandaran
No full textThe timing and process of Neolithisation in Eastern Mazandaran have been a topic of debate among
archaeologists, as the Southern Caspian littoral is regarded as a likely route for the spread of Neolithic culture to
Central Asia. Until recently, there was no reliable evidence connecting this region to the Pottery Neolithic (PN)
sites of Central Asia (Djeitun Culture). However, an archaeological field survey conducted in 2021 at the site of
Tappeh Fakhi has provided direct evidence of Djeitun pottery in this region. Since the processes of endogenous
or exogenous Neolithisation and domestication in Eastern Mazandaran remain controversial, sites such as Tappeh
Fakhi can shed light on the issues we are facing. When compared to adjacent regions such as the Central Plateau,
Shahrud Plain, Gorgan Plain, Khorasan Region, and Western Central Asia, the Neolithic sherds found at Tappeh
Fakhi suggest a chronology beginning in the final 7th millennium BCE. In contrast, the Chalcolithic sherds indicate
that the site continued to be occupied until the end of the 5th millennium BCE. During the field survey, two potential
paths for the introduction of the Djeitun culture to Eastern Mazandaran were considered: one from the southern
slopes (Shahrud region) to the northern part of the Alborz mountains, and another through the Gorgan plain. Given
the location of Tappeh Fakhi, it appears to be a significant site that connects the two cultures, i.e. the Djeitun and
Caspian Neolithic Software, likely through the Gorgan Plain. This paper will summarise the known aspects of the
PN in Eastern Mazandaran and its inter-regional interactions
Credit and climate risk in green debt markets: corporate bonds and energy efficient securitisation
Get PDFThis thesis explores how environmental information affects credit risk and pricing in fixed-income markets. It focuses on three areas: corporate green bonds, Green RMBS, and
mortgage loans with energy performance data. The aim is to understand whether markets
respond to environmental signals, and under what conditions these signals matter most.
The first empirical investigation examines corporate green bonds in the secondary market.
The analysis shows that the negative yield differential of green over conventional bonds
is dynamic and reacts to climate policy events. Certified green bonds in environmentally
material industries enjoy a larger premium, while non-certified issues often face scepticism
over greenwashing. Certified bonds also maintain narrower spreads during natural disasters
and heightened media coverage of climate change.
The second empirical investigation studies Green Residential Mortgage-Backed Securities
(RMBS) deals. Using detailed loan and tranche data, it finds that deals labelled as green are
backed by loans with lower delinquency risk and are more likely to produce investment-grade
tranches. Moreover, stress simulations confirm that green-labelled structures absorb smaller
losses under adverse conditions.
The third empirical investigation shifts focus to property-level energy efficiency. It
harmonises Energy Performance Certificate (EPC) data from multiple countries to test
whether buildings’ energy performance predicts loan-level delinquency. The results show
that less efficient homes are associated with higher arrears and default risk, especially
for lower-income borrowers and during periods of high energy inflation. These findings
underline the importance of energy efficiency for loan-level credit risk assessment and its
relevance to EU goals on the green transition and energy security.
Overall, the thesis shows that environmental features can enhance markets’ assessment
of risk, with three key insights: environmental factors are financially relevant, credibility is
critical, and effects are context-dependent
Tribal epistemologies and the discursive construction of COVID-19 knowledge
Get PDFThis special issue aims to explore questions of how knowledge is dynamically produced through discourse and the role that “knowledge-in-action” plays in developing and maintaining identities and group allegiances (Moje 2011) in the context of the COVID-19 pandemic. A wide range of practices in which epistemological conflicts and incongruities were implicated are addressed, including mask wearing, naming practices, the representation of scientific knowledge to lay-people, intercultural (mis)communication, nationalism, and online practices of argumentation. The papers focus not just on the relationship between knowledge and ways of representing it (Fairclough 2000; Lemke 1995), but on how ways of knowing unfold in and drive interactions between institutions, communities and individuals, opening up and closing off routes to identification and belonging. All of these papers come from a consortium of discourse analysts from the UK and Hong Kong who worked together on issues related to COVID-19 during the pandemic (see https://viraldiscourse.com)
Laughing in the face of the law: humour as a thermostat activating social change for porn workers
Get PDFThe vulnerability of sex workers in the porn industry is a heated debate within feminism. The UK 2014 Audiovisual Media Services Regulations and 2017 Digital Economy Act, which burden the production of online pornography, provoked sex workers’ Face-Sitting and Kink Olympixxx protests. This paper investigates how throughout these protests, humour communicates sex workers’ discomfort on this legislation. Arguing that humour is a thermostat that senses public uneasiness and slowly activates social change, this paper examines the two protests highlighting how sex workers employed unrefined bawdy humour to unearth their neglected rights and move towards more adequate rights
Lexico‐semantic attrition of native language: evidence from Russian–Hebrew bilinguals
No full textNative language (L1) attrition is ubiquitous in modern globalized society, but its cognitive/psycholinguistic mechanisms are poorly understood. We investigated lexico‐semantic L1 attrition in L1 Russian immigrants in Israel, who predominantly use their second language (L2), Hebrew, in daily life. We included Russian monolinguals as a control group. We tested two potential causal mechanisms of attrition: L2 interference versus L1 disuse. Participants completed a fill‐the‐gap task in two conditions: accuracy (producing one exactly matching word) and scope (providing as many synonyms as possible). We expected L2 interference and L1 disuse to lead to the differential reduction of accuracy and scope features, respectively. Lower scores for attriters emerged in the accuracy but not in the scope condition. Moreover, attitude towards L1 influenced attriters’ accuracy—but not scope—performance, with higher L1 preference predicting higher accuracy. We provide evidence for lexico‐semantic attrition in adult immigrants, pointing to L2 interference as the primary cause of impaired lexical retrieval
From overlooked objects to digital ‘icons’: evaluating the role of social media in exhibition making and the creation of more participatory and democratic museums
Get PDFThis article highlights the opportunity for social media to play a greater role in critically and democratically evaluating museum displays and exhibitions in order to increase their popular appeal. By comparing datasets obtained from traditional (often in-person) evaluation techniques with those expressed by visitors using social media platforms, it is possible to assess the relative strengths and weaknesses of each, resulting in a more holistic approach to audience reception. Based on the results of a curatorial experiment to diversify the type of objects displayed in a major exhibition held at the British Museum (‘The world of Stonehenge’), this paper uses qualatitive and quantitative (over 1000 images and 50,000 words) methods to explore the potential, wider, role of visitor-generated social media content. We highlight the potential of this approach for critiquing and rethinking exhibition design, marketing and the ethics that underpin pressing moral issues such as the display of human remains and the reductive and problematic process of ‘iconification’ whereby certain museum objects and heritage assets are selected to represent whole geographies, peoples and eras
The cloud people of Naupallacta: Middle Horizon Camelid management in the hinterland of the Chicha Soras Valley
Get PDFArchaeological, ethnographic, and rural development work over the past 50 years in the Chicha Soras area has revealed that the region was first intensively used from the Middle Horizon Epoch 2 (AD 680/700–900) onward. An introduced population here had its seven occupation sites focused on the southern part of the Chicha valley and supported agriculture and textile production, while settlement in the upper Rio Yanamayo and west of the Rio Huayllaripa comprising the Naupallacta complex, served the management of camelid herds, which provided transport, fiber, and meat resources. The interrelationship between the arable farmers and camelid pastoralists produced a mutually supportive resource base for the local agropastoral community, and tribute produced for distant elite Wari administrators. This is confirmed by the interdisciplinary data sets presented here, including demographic, architectural, artefactual, archaeozoological, paleoenvironmental and aDNA results