Usage

When downloading, specify which files you want to download and set the split to train (required by datasets). from datasets import load_dataset

nouns = load_dataset("fairnlp/holistic-bias", data_files=["nouns.csv"], split="train") sentences = load_dataset("fairnlp/holistic-bias", data_files=["sentences.csv"], split="train")

  Dataset Card for Holistic Bias

This dataset contains the source data of the Holistic Bias dataset as described by Smith et. al. (2022).… See the full description on the dataset page: https://huggingface.co/datasets/fairnlp/holistic-bias.

With the widespread use of knowledge graphs (KG) in various automated AI systems and applications, it is very important to ensure that information retrieval algorithms leveraging them are free from societal biases. Previous works have depicted biases that persist in KGs, as well as employed several metrics for measuring the biases. However, such studies lack the systematic exploration of the sensitivity of the bias measurements, through varying sources of data, or the embedding algorithms used. To address this research gap, in this work, we present a holistic analysis of bias measurement on the knowledge graph. First, we attempt to reveal data biases that surface in Wikidata for thirteen different demographics selected from seven continents. Next, we attempt to unfold the variance in the detection of biases by two different knowledge graph embedding algorithms - TransE and ComplEx. We conduct our extensive experiments on a large number of occupations sampled from the thirteen demographics with respect to the sensitive attribute, i.e., gender. Our results show that the inherent data bias that persists in KG can be altered by specific algorithm bias as incorporated by KG embedding learning algorithms. Further, we show that the choice of the state-of-the-art KG embedding algorithm has a strong impact on the ranking of biased occupations irrespective of gender. We observe that the similarity of the biased occupations across demographics is minimal which reflects the socio-cultural differences around the globe. We believe that this full-scale audit of the bias measurement pipeline will raise awareness among the community while deriving insights related to design choices of data and algorithms both and refrain from the popular dogma of ``one-size-fits-all''.

BackgroundDespite a well-recognised burden of disabling physical symptoms compounded by co-morbidities, psychological distress and social isolation, the needs of people with severe chronic obstructive pulmonary disease (COPD) are typically poorly addressed. AimTo assess the effectiveness of interventions designed to deliver holistic care for people with severe COPD. MethodsWe searched 11 biomedical databases, three trial repositories (January 1990-March 2012; no language restrictions) and contacted international experts to locate published, unpublished and in-progress randomised controlled trials (RCTs), quasi-RCTs and controlled clinical trials (CCTs) that investigated holistic interventions to support patients with severe COPD in any healthcare context. The primary outcome was health-related quality of life (HRQoL). Quality assessment and data extraction followed Cochrane Collaboration methodology. We used a piloted data extraction sheet and undertook narrative synthesis. ResultsFrom 2,866 potentially relevant papers, we identified three trials: two RCTs (from United States and Australia), and one CCT (from Thailand): total 216 patients. Risk of bias was assessed as moderate in two studies and high in the third. All the interventions were led by nurses acting in a co-ordinating role (e.g. facilitating community support in Thailand, providing case-management in the USA, or co-ordinating inpatient care in Australia). HRQoL improved significantly in the Thai CCT compared to the (very limited) usual care (p

We studied differences in localizing the centres of flow in radially expanding and contracting patterns in different regions of the visual field. Our results suggest that the perceived centre of a peripherally viewed expanding pattern is shifted towards the fovea relative to that of a contracting pattern, but only in the lower right and upper right visual quadrants and when a single speed gradient with appropriate overall speeds of the trajectories of the moving dots was used. The biases were not systematically related to differences of sensitivity to optic flow in different quadrants. Further experiments demonstrated that the biases were likely due to a combination of two effects: an advantage of global processing in favor of the lower visual hemifield and a hemispheric asymmetry in attentional allocation in favor of motion-induced spatial displacement in the right visual hemifield. The bias in the lower right visual quadrant was speed gradient-sensitive and could be reduced to a non-significant level with the usage of multiple speed gradients, possibly due to a special role of the lower visual hemifield in extracting global information from the multiple speed gradients. A holistic processing on multiple speed gradients, rather than a predominant processing on a single speed gradient, was likely adopted. In contrast, the perceived bias in the upper right visual quadrant was overall speed-sensitive and could be reduced to a non-significant level with the reduction of the overall speeds of the trajectories. The implications of these results for understanding motion-induced spatial illusions are discussed.

Retirement is a pivotal life transition that often changes routines, identity, and objectives. With increasing life expectancies and evolving societal norms, examining the interplay between retirement anxiety and life satisfaction is vital. This study delves into this relationship, recognising the complexities of retirement. A systematic review and meta-analysis followed PRISMA guidelines. Research from 2003 to 2023 was sourced from databases like CINAHL, PubMed/Medline, PsycINFO, ERIC, and Google Scholar, focusing on diverse methodologies and outcomes related to retirement registered in Prospero database (CRD42023427949). The quality assessment used an eight-criterion risk of bias scale, and analyses included qualitative and quantitative approaches, such as random-effects meta-analysis and moderator analyses. After reviewing 19 studies with varied geographical and demographic scopes, a mixed relationship between retirement and life satisfaction emerged: 32% of studies reported a positive relationship, 47% were negative, and 21% found no significant correlation. Meta-analysis indicated high heterogeneity and non-significant mean effect size, suggesting no consistent impact of retirement on life satisfaction. Moderator analyses highlighted the influence of measurement tools on outcomes. The findings reveal a complex interplay between retirement anxiety and life satisfaction, stressing the need for holistic retirement policies that encompass mental health, social integration, and adaptability, focusing on cultural sensitivity. Challenges include potential biases in data sources, methodological diversity, the scarcity of longitudinal studies, and difficulties in addressing recent societal shifts, like the COVID-19 pandemic. Variability in measurement tools and possible publication bias may have also influenced results. This study contributes to understanding retirement, emphasising the relationship between retirement anxiety and life satisfaction. It advocates for ongoing, detailed, culturally informed research to grasp retirement’s multifaceted aspects fully.

A survey targeting employees in the energy sector was conducted in summer 2023. The survey was designed to provide insights into company culture, gender biases and their impact on career decisions across traditional, renewable and energy transition sectors. With the demand for professionals in the energy industry expected to increase further, it is crucial to offer attractive working conditions to avoid skills shortages. Our study focuses on assessing employee satisfaction to identify opportunities for improvement. What sets the study apart is its comprehensive approach. We examine not only organisational culture and subtle gender bias, but also general workplace satisfaction as well as misconduct (such as bullying and discrimination), and how these factors influence turnover intentions. By considering these interrelated elements, we hope to gain a holistic understanding of the challenges faced by professionals in the energy sector.

The questionnaires were implemented on the web, namely on EUSurvey and Netigate. The survey made use of a structured questionnaire with a range of closed questions and the option to add open text responses where appropriate. The survey was distributed with the help of the broad network of organisations and experts in the studied countries. In addition, an industry panel provided by Netigate was used to reach the target number of responses. The survey was aimed an any person in the target population of people working in the energy industry, irrespective of gender. The survey took 15 to 20 minutes to answer. All answers were anonymous, and the respondents were not asked any identifiable information, such as e-mail address, to avoid bias. The questionnaire was available in the primary national language of each EU27 country and special attention was paid to formulate the questions in a non-biased, clear and gender-neutral manner, avoiding jargon or ambiguity. No weighting was applied to the dataset.

Joanna Briggs Institute (JBI) risk of bias assessment for Randomized Controlled Trials (RCT).

Feature importance and relations to Grid-Group Theory’s cultural bias.

Feature recommendations for in-vehicle flash flood app based on Grid-Group Cultural Theory.

Flash floods are severe disaster that have caused enormous damage to people, property, and the environment. Despite the conventional emphasis on technical and engineering solutions in controlling flash flood disasters, this study investigates the understudied issue of user-centric cultural viewpoints, inspired by Grid-Group Cultural Theory, and their potential impact on crisis management. The study collected 351 responses, primarily targeting adults in flood-prone areas using convenience sampling method with the goal of exploring cultural bias for feature identification of in-vehicle flash flood app. Accordingly, the research investigates the participants responses using quantitative approach which includes descriptive statistics, exploratory factor analysis, average factor, and rank scoring analysis to uncover critical user-centric cultural traits that might improve preparedness, response, and recovery activities during flash flood disasters. The findings of the study identified distinct cultural biases that impact perceptions and preferences regarding features of an in-vehicle flash flood app. By integrating Grid-Group Cultural Theory as a framework for analysis, the study highlights the importance of incorporating diverse cultural perspectives into flash flood management strategies. The result emphasizes the need to apply a holistic approach that integrates people’s knowledge and practices with technical solutions. Recommendations of features for future development of in-vehicle flash flood app is provided based on each cultural bias aligned with the theory to build more resilient communities in the face of flash flood occurrences.

Fueled by early success stories, the neuromarketing domain advanced rapidly during the last 10 years. As exciting new techniques were being adapted from medical research to the commercial domain, many neuroscientists and marketing practitioners have taken the chance to exploit them so as to uncover the answers of the most important marketing questions. Among the available neuroimaging technologies, electroencephalography (EEG) stands out as the less invasive and most affordable method. While not equally precise as other neuroimaging technologies in terms of spatial resolution, it can capture brain activity almost at the speed of cognition. Hence, EEG constitutes a favorable candidate for recording and subsequently decoding the consumers' brain activity. However, despite its wide use in neuromarketing, it cannot provide the complete picture alone. In order to overcome the limitations imposed by a single monitoring method, researchers focus on more holistic approaches. The exploitation of hybrid EEG schemes (e.g., combining EEG with eye-tracking, electrodermal activity, heart rate, and/or other) is ever growing and will hopefully allow neuromarketing to uncover consumers' behavior. Our survey revolves around last-decade hybrid neuromarketing schemes that involve EEG as the dominant modality. Beyond covering the relevant literature and state-of-the-art findings, we also provide future directions on the field, present the limitations that accompany each of the commonly employed monitoring methods and briefly discuss the omni-present ethical scepticizm related to neuromarketing.

This study explores the factors influencing college students’ acceptance and resistance toward generative AI technologies by integrating three theoretical frameworks: the Technology Acceptance Model (TAM), Protection Motivation Theory (PMT), and Social Exchange Theory (SET). Using data from 407 respondents collected through a structured survey, the study employed Structural Equation Modeling (SEM) to examine how functional factors (perceived usefulness, ease of use, and reliability), risk factors (privacy concerns, data security, and ethical issues), and sociolegal factors (trust in governance and regulatory frameworks) impact user attitudes. Results revealed that functional factors significantly enhanced acceptance while reducing resistance, whereas risk factors amplified resistance and negatively influenced acceptance. Sociolegal factors emerged as critical mediators, mitigating the negative impact of perceived risks and reinforcing the positive effects of functional perceptions. The study responds to prior feedback by offering a more integrated theoretical framework, clearly articulating how TAM, PMT, and SET interact to shape user behavior. It also acknowledges the limitations of using a student sample and discusses the broader applicability of the findings to other demographics, such as professionals and non-academic users. Additionally, the manuscript now highlights demographic diversity, including variations in age, gender, and academic discipline, as relevant to AI adoption patterns. Ethical concerns, including algorithmic bias, data ownership, and the labor market impact of AI, are addressed to offer a more holistic understanding of resistance behavior. Policy implications have been expanded with actionable recommendations such as AI bias mitigation strategies, clearer data ownership protections, and workforce reskilling programs. The study also compares global regulatory frameworks like the GDPR and the U.S. AI Bill of Rights, reinforcing its practical relevance. Furthermore, it emphasizes that user attitudes toward AI are dynamic and likely to evolve, suggesting the need for longitudinal studies to capture behavioral adaptation over time. By bridging theory and practice, this research contributes to the growing discourse on responsible and equitable AI adoption in higher education, offering valuable insights for developers, policymakers, and academic institutions aiming to foster ethical and inclusive technology integration.