Revenue management has been applied to the restaurant industry, but restaurant operators have been disinclined to apply various types of RM approaches, due to apprehension for customer's possible expressions of dissatisfaction. To relieve this reluctance, restaurant operators may need to understand how their customers perceive capacity limitations. While customers are more familiar with RM practices in traditional RM industries (e.g., airlines or hotels) with fixed capacities, perceptions of capacity limitations in restaurants (relatively flexible capacity) may influence customers' perceptions of RM practices. In addition, the price difference between high-demand periods and low-demand periods may have differential impacts on customers' perceptions of value of the restaurant's expected offering and the fairness of RM practices. Based on commodity theory and equity theory, this study hypothesizes that two main effects, perceived scarcity of space in a restaurant and price differences between high-demand and low-demand periods, influence perceived value of a restaurant's offering and fairness perceptions of a restaurant's RM practice. As hypothesized, the negative effects of price difference on fairness perceptions are supported by the results, but the effect on perceived value has support only from the results of structural equation modeling. Unexpectedly, the main effect of perceived scarcity of space does not influence either perceived value of a restaurant's expected offering or fairness perceptions for a restaurant's RM practice. Interesting results arose found from supplementary analyses and suggest future research directions. / Tourism and Sport

<p dir="ltr">Data pipelines are core machine learning components essential for moving data through various stages and applying transformations to enhance data quality for model training, thereby improving performance and efficiency. However, as data volumes grow, optimizing these pipelines becomes increasingly complex, which can impact performance and increase the costs of finding the optimal pipeline. Data-centric systems are found across various sectors, including finance, education, marketing, and healthcare, which are trained on historical data. After that, systems need to be monitored, and continuous testing is required to ensure the performance of new incoming data. However, when the system encounters failures with new incoming data, debugging is needed to find the data point that is causing the system to fail. Finding the optimal pipeline for new data can also be daunting. In this research, we aim to address these challenges by proposing an approach that uses the GRASP method to find the new pipeline and a data profile to find the cause of the disconnect between the pipeline and data.</p>

This thesis investigates the relationship between generalised trust, heterogeneity and fairness in the EU. High levels of generalised trust have been found to be linked to economic growth and lower transaction costs, while heterogeneous populations have been found to be linked to lower levels of trust. As the world is diversifying on virtually all fronts, a negative relationship between heterogeneity and generalised trust could have dire implications. Other empirical studies have found that heterogenety loses its importance when accounting for societal fairness, speaking to the fact that a fair and democratic society is what determines generalised trust, rather than inter-group differences. But as no such studies have been conducted on contemporary data, it is unknown how the relationship between generalised trust, heterogeneity and fairness holds today. To investigate this relationship, we conducted two-step hierarchical logit regressions on contemporary data, including over 40.000 observations from 23 EU countries. To capture contemporary heterogeneity, we constructed our own fractionalisation indices on this data, as the fractionalisation in- dices most commonly used in previous studies were constructed on data from the 1990s and early 2000s. Our results indicate that both fairness and heterogeneity are significant predictors for generalised trust in the EU. We theorise that this might be because the importance of heterogeneity for generalised trust is context-dependent, only mattering once countries have established a certain level of fairness.

The distributed and privacy sensitive nature of cellular networks make them strong candidates for optimization using Federated Learning, but this exposes them to a problem inherent to the learning paradigm: performance inequality due to heterogeneous client data distributions. The prevailing approach of enforcing uniform client performance ignores client-specific performance limitations due to different levels of irreducible uncertainty present in their data, resulting in deteriorated network performance. To address this issue, this thesis introduces two novel federated algorithms designed to enhance learning efficiency and ensure fairness in the presence of heteroscedastic noise, reflecting the distributive justice principles of utilitarianism and equality. Under these circumstances, the proposed algorithms are shown to significantly improve overall performance and performance fairness. The deployment of these algorithms promises a dual benefit: enhancement in network performance and a fairer distribution of service quality for end users.

Social media sites have been perceived as a "common digital town square" in which opinions are exchanged at an unprecedented scale and speed. People not only share their own opinions on controversial issues but also consume news and assimilate opinions shared by friends in their social circles. The process by which individuals update their opinions in a social network is called opinion dynamics. This dissertation focuses on the negative consequences that arise from opinion dynamics, namely unfairness, polarization, and radicalization. We leverage techniques from social network modeling, graph theory, and supervised machine learning to formalize the notions of fairness in opinion dynamics over social networks, diagnose which and when algorithms exacerbate unfairness in networks, and design mitigation algorithms to counter radicalization. In the first project, we formalize two aspects of fairness in opinion dynamics, namely procedural fairness and distributive fairness. Through theoretical analysis and simulations on real-world social networks, we show how the combined effect of homophilous and reinforcing dynamics plays a special role in both types of fairness. In the second project, we study radicalization pathways that lead users from moderate to more extreme content and propose algorithms to mitigate radicalization. In particular, we study and propose the concept of gateway entities, i.e., non-problematic entities that are nevertheless associated with a higher likelihood of future engagement with radicalized content. We show, via a real-world application on Facebook groups, that a simple definition of gateway entities can be leveraged to reduce exposure to radicalized content without adversely impacting user engagement metrics. Through offline experiments, we show that survival analysis-based methods are effective at identifying individuals at risk. In the third project, we study the interplay between algorithms over social networks and fairness in opinion dynamics. We propose a framework that allows us to study the joint effects of algorithms over social networks and the opinion dynamic process. Through extensive simulations, we demonstrate that people-recommender algorithms do not exacerbate procedure unfairness but have a significant impact on distributive fairness, and polarization reduction algorithms have no significant impact on fairness. We close by discussing the limitations of our work and directions for future research.

Yes / We provide a comprehensive but critical review of research on applicant reactions to selection procedures published since 2000 (n = 145), when the last major review article on applicant reactions appeared in the Journal of Management. We start by addressing the main criticisms levied against the field to determine whether applicant reactions matter to individuals and employers (“So what?”). This is followed by a consideration of “What’s new?” by conducting a comprehensive and detailed review of applicant reaction research centered upon four areas of growth: expansion of the theoretical lens, incorporation of new technology in the selection arena, internationalization of applicant reactions research, and emerging boundary conditions. Our final section focuses on “Where to next?” and offers an updated and integrated conceptual model of applicant reactions, four key challenges, and eight specific future research questions. Our conclusion is that the field demonstrates stronger research designs, with studies incorporating greater control, broader constructs, and multiple time points. There is also solid evidence that applicant reactions have significant and meaningful effects on attitudes, intentions, and behaviors. At the same time, we identify some remaining gaps in the literature and a number of critical questions that remain to be explored, particularly in light of technological and societal changes. / Research grant from the Social Sciences and Humanities Research Council of Canada awarded to Julie M. McCarthy (No. 435-2015-0220).

Yes / This paper explores the issue of fairness in global supply chains. Taking the Western European clothing supply chain as a case study, we demonstrate how applying a normative indicator in Social Life Cycle Assessment (SLCA) can contribute academic and practical insights into debates on fairness. To do so, we develop a new indicator that addresses some of the limitations of the living wage for SLCA. We extend the standard form of living wage available for developing countries to include income tax and social security contributions. We call this extension 'living labour compensation'. Using publically available data, we estimate net living wages, gross living wages, and living labour compensation rates for Brazil, Russia, India, and China (BRIC) in 2005. We then integrate living labour compensation rates into an input-output framework, which we use to compare living labour compensation and actual labour compensation in the BRIC countries in the Western European clothing supply chain in 2005. We find that in 2005, actual labour compensation in the Western European clothing supply chain was around half of the living labour compensation level, with the greatest difference being in the Agricultural sector. Therefore, we argue that BRIC pay in the Western European clothing supply chain was unfair. Furthermore, our living labour compensation estimates for BRIC in 2005 are ~ 35% higher than standard living wage estimates. Indeed, adding income taxes and employee social security contributions alone increases the living wage by ~ 10%. Consequently, we argue there is a risk that investigations based on living wages are not using a representative measure of fairness from the employee's perspective and are substantially underestimating the cost of living wages from an employer's perspective. Finally, we discuss implications for retailers and living wage advocacy groups. Living labour compensation extends the living wage, maintaining its strengths and addressing key weaknesses. It can be estimated for multiple countries from publically available data and can be applied in an input-output framework. Therefore, it is able to provide a normative assessment of fairness in complex global supply chains. Applying it to the Western European clothing supply chain, we were able to show that pay for workers in Brazil, Russia, India, and China is unfair, and draw substantive conclusions for practice.

No / In this chapter we use global multi-regional input-output analysis to explore how globalisation has impacted fairness along Western European clothing supply chains. Our analysis shows that while globalisation has made the Western European clothing supply chain ‘fairer’ by increasing employment opportunities and income for workers in Brazil, Russia, India and China (BRIC), it has failed to make the supply chain fair. Despite large increases in the labour compensation received by BRIC workers in the Western European clothing supply chain, labour compensation is still insufficient to support a decent standard of living and cannot, therefore, be considered fair.

The use of artificial intelligence (AI) tools is rapidly increasing in organizations worldwide. The purpose of this cross-sectional experimental study is to compare perceptions of procedural fairness and perceptions of discrimination based on sexual orientation in AI-assisted versus human recruitment processes. Using vignettes, we asked participants (N = 278) to assess recruitment processes and fictional applicant’s LinkedIn profiles where sexual orientation was signaled. In more detail, after participants were informed about the negative decision made either by the AI tool or the human recruitment team, they were asked to report their perceptions of procedural fairness and perceptions of discrimination based on sexual orientation. We used independent samples t-tests and two-way ANOVA to analyze our main hypotheses. Our findings reveal that AI-assisted recruitment processes are perceived as less procedurally fair than human recruitment processes. In contrast, the results indicate that AI-assisted recruitment processes are perceived as less discriminatory for non-heterosexual applicants compared to heterosexual applicants. The findings cover a gap in research on perceived discrimination based on sexual orientation in AI-assisted recruitment. Moreover, by shedding light on the complexities of perceptions concerning AI-assisted and human recruitment processes, our findings underline the emerging need for organizations to invest in AI literacy, increase employees’ AI familiarity, and openly commit to AI legislation. Lastly, our findings may provide insights for informing talent acquisition strategies, learning and development programs, and diversity, equity, and inclusion initiatives in digitized companies. / <p>The final thesis submission was on the 02.06.2024 however, the final seminar including the thesis defence was on the 21.05.2024.</p><p>Please find the datasets (raw and cleaned) and all further supplementary materials on OSF, these are hyperlinked in the final document.</p>

This study introduces an effective method to address the generalization challenge in pretrained language models (PLMs), which affects their performance on diverse linguistic data beyond their training scope. Improving PLMs' adaptability to out-of-distribution (OOD) data is essential for their reliability and practical utility in real-world applications. Furthermore, we address the ethical imperative of fairness in PLMs, particularly as they become integral to decision-making in sensitive societal sectors. We introduce gender-tuning, to identify and disrupt gender-related biases in training data. This method perturbs gendered terms, replacing them to break associations with other words. Gender-tuning stands as a practical, ethical intervention against gender bias in PLMs. Finally, we present FairAgent, a novel framework designed to imbue small language models (SLMs) with fairness, drawing on the knowledge of large language models (LLMs) without incurring the latter's computational costs. FairAgent operates by enabling SLMs to consult with LLMs, harnessing their vast knowledge to guide the generation of less biased content. This dynamic system not only detects bias in SLM responses but also generates prompts to correct it, accumulating effective prompts for future use. Over time, SLMs become increasingly adept at producing fair responses, enhancing both computational efficiency and fairness in AI-driven interactions.