RESUMO
Recommender systems (RSs) have become an integral part of the hiring process, be it via job advertisement ranking systems (job recommenders) for the potential employee or candidate ranking systems (candidate recommenders) for the employer. As seen in other domains, RSs are prone to harmful biases, unfair algorithmic behavior, and even discrimination in a legal sense. Some cases, such as salary equity in regards to gender (gender pay gap), stereotypical job perceptions along gendered lines, or biases toward other subgroups sharing specific characteristics in candidate recommenders, can have profound ethical and legal implications. In this survey, we discuss the current state of fairness research considering the fairness definitions (e.g., demographic parity and equal opportunity) used in recruitment-related RSs (RRSs). We investigate from a technical perspective the approaches to improve fairness, like synthetic data generation, adversarial training, protected subgroup distributional constraints, and post-hoc re-ranking. Thereafter, from a legal perspective, we contrast the fairness definitions and the effects of the aforementioned approaches with existing EU and US law requirements for employment and occupation, and second, we ascertain whether and to what extent EU and US law permits such approaches to improve fairness. We finally discuss the advances that RSs have made in terms of fairness in the recruitment domain, compare them with those made in other domains, and outline existing open challenges.
RESUMO
Natural language processing (NLP) is a rapidly evolving field at the intersection of linguistics, computer science, and artificial intelligence, which is concerned with developing methods to process and generate language at scale. Modern NLP tools have the potential to support humanitarian action at multiple stages of the humanitarian response cycle. Both internal reports, secondary text data (e.g., social media data, news media articles, or interviews with affected individuals), and external-facing documents like Humanitarian Needs Overviews (HNOs) encode information relevant to monitoring, anticipating, or responding to humanitarian crises. Yet, lack of awareness of the concrete opportunities offered by state-of-the-art techniques, as well as constraints posed by resource scarcity, limit adoption of NLP tools in the humanitarian sector. This paper provides a pragmatically-minded primer to the emerging field of humanitarian NLP, reviewing existing initiatives in the space of humanitarian NLP, highlighting potentially impactful applications of NLP in the humanitarian sector, and describing criteria, challenges, and potential solutions for large-scale adoption. In addition, as one of the main bottlenecks is the lack of data and standards for this domain, we present recent initiatives (the DEEP and HumSet) which are directly aimed at addressing these gaps. With this work, we hope to motivate humanitarians and NLP experts to create long-term impact-driven synergies and to co-develop an ambitious roadmap for the field.
