The field of human resources (HR) encompasses a wide range of tasks where the application of natural language processing (NLP) holds significant promise. Applications such as talent acquisition, career development guidance, performance management, and ongoing education and training involve a substantial amount of unstructured, semi-structured, and structured data. NLP can offer (semi-)automated solutions to address the challenges associated with these tasks. At the same time, the integration of AI in HR applications also presents certain risks and concerns, such as fairness, privacy, reproducibility, controllability, and transparency. There is an enormous opportunity to leverage advanced NLP techniques to address these challenges.
Scope
The NLP4HR workshop focuses on the application of NLP in HR, covering a wide range of topics such as:
- Acquiring HR-specific knowledge from a variety of sources
- Information extraction from HR documents, including job descriptions and resumes
- Representation learning for HR entities like jobs, job seekers, and employers
- Analysis of opinions on companies
- Search and recommendation systems for recruiters and job seekers
- Conversational HR assistants
- Generation of HR-related documents
- Question answering systems for HR-related inquiries
- Identification and rectification of biases
These areas intersect not only with established NLP tasks but also with areas like data mining and management, inspiring workshops across various communities. These include the following: RecSys in HR (at RecSys), Talent and Management Computing (at KDD), and AI for Human Resources and Public Employment Services (at ECML). The NLP4HR workshop particularly aims to bring together NLP researchers to advance innovative solutions for HR challenges. To name a few related challenges:
Parsing: The extraction of key details from documents such as job descriptions and resumes is vital. This includes job responsibilities, required qualifications, and benefits in job descriptions, plus skill sets, educational backgrounds, and certifications in resumes. One of the representative tasks, skill identification—despite appearing as a standard information extraction task—presents unique difficulties in HR documents due to issues like a large number of target labels and the need to identify nuanced distinctions (e.g., soft vs hard skills).
Person-Job Fit: Arguably, matching job openings with applicants stands as the most important goal of HR technologies. Recommending jobs to seekers, and vice versa might seem like a straightforward recommendation task. However, job matching is a uniquely complex challenge for several reasons. Deep comprehension of semantic elements, including job duties and applicants’ skills/experience, has become increasingly crucial in recruiting. This has led to the development of skill-based matching systems. Furthermore, explaining matching results in natural language is an essential task where NLP techniques can provide valuable assistance. It’s particularly interesting to explore how large language models can be leveraged to assist recruiters/job seekers in decision-making, as already explored in recommender systems research.
Writing Support: Crafting HR documents often necessitates a deep understanding of the job, company, and market trends. Therefore, the process of writing job descriptions often necessitates close discussions with domain experts. Crafting compelling resumes or cover letters can be challenging for job seekers, as it’s often unclear what recruiters are looking for. Several prior works have tackled this problem (Liu et al., 2020; Qin et al., 2023, among others). Recently, the performance of language models has greatly facilitated automatic writing support at production quality (e.g., Indeed, LinkedIn, Grammarly).
Knowledge Discovery: Extracted information can have significant implications for social and economic studies. For instance, analytic reports by domain experts can provide valuable insights for economic and political decision-making. Prior work has demonstrated that machine learning models can offer insights into the valuation of various skills.
Bias and Fairness: On the flip side, the use of automatic systems can inadvertently pick undesired patterns that negatively impact human users. Previous studies have highlighted biased associations between gender and occupation and between names (which encode attributes such as gender and nationality) and occupation within NLP systems. Techniques for mitigating such biases have been extensively studied and can enhance applications in the HR domain.
Highlights
The workshop is designed to encourage lively discussions and will feature prominent researchers and technical leaders from academia and industry who will identify issues and propose new directions.
Keynotes
The workshop will feature distinguished invited speakers who bring expertise from academia, the OECD, and industry. They will cover a range of critical topics:
- Barbara Plank from the University of Munich (LMU) will delve into resumes, job descriptions, skills extraction, and matching. In particular, Plank will look at benchmarks, models, and approaches as they relate to HR.
- Marko Grobelnik, from OECD.AI and the Jožef Stefan Institute, will introduce the OECD/UNESCO program on AI in work, innovation, productivity, and he’ll participate in a skills panel discussion as well.
- Trey Causey from Indeed will discuss ethical implications of technologies in the HR domain.
- David Graus from Randstad and University of Amsterdam will present his work on NLP-driven recommendation systems in HR.
The panel, tentatively titled “HR in the era of large pre-trained models: sorting out ‘the good, the bad, and the ugly,” will shed light on the impacts of AI-based approaches in HR tasks and whether organizations are actively addressing biases in their models and systems.
Join Our Inaugural NLP4HR Workshop
We invite scholars, researchers, and practitioners to contribute their expertise and insights to the inaugural NLP4HR workshop. We are excitedly anticipating an eclectic mix of submissions covering a broad range of HR-related topics. See our call for papers section on our website, if you are interested in submitting your research. We offer mentorship opportunities for those interested in a mentor through the submission process and for professionals who’d like to make an impact on a young professional via mentorship. For more information, please visit our workshop website. Should you have any questions or concerns, feel free to reach out to the organizers at nlp4hr-workshop@megagon.ai.
Written by: Estevam Hruschka, Naoki Otani, and Megagon Labs.