Machine Learning for Post-Diploma Educational and Career Guidance: A Scoping Review in AI-Driven Decision Support Systems

Flavio Manganello^1*Elisa Rasca²Alberto Villa²Andrea Maddalena¹Giannangelo Boccuzzi¹

• ¹Institute of Didactic Technologies, Department of Human and Social Sciences, Cultural Heritage, National Research Council (CNR), Genova, Italy
• ²Hybrid Intelligence, Capgemini (Italy), Rome, Lazio, Italy

The increasing complexity of career decision-making, shaped by rapid technological advancements and evolving job markets, underscoreshighlights the need for more effectiveresponsive and datainformed post-diploma guidance. ArtificialMachine learning (ML), a core component of artificial intelligence has gained significant, is gaining attention in this domain, offering new possibilities for its potential to support personalized educational and career support. Among artificial intelligence techniques, machine learning stands out as a key approach, enabling data-driven decisionmakingdecisions by analyzing academic records, personalindividual preferences, and labor market trends. Howeverdata. Despite growing interest, research in this areafield remains fragmented, requiring a systematic examination of existing studies. and methodologically diverse. This scoping review maps the application of machine learningML in post-diploma guidance, analyzing by examining the types of models used, data sources, reported outcomes, and ethical considerations related to fairness, privacy, and transparency. A systematic search was conducted in of Scopus and Web of Science was conducted, with the final search completed on December 31, 2023. Only peerreviewed journal articles, conference proceedings, and book chapters focusing on machine learning in career or educational guidance were included. A total of 183 records were retrieved, of which 21Twenty-one studies met the inclusion criteria. Most studies applied, primarily employing supervised or mixed-method machine learningML techniques, primarily for university and career to develop recommendation systems. or predictive models. While several contributions report positive technical performance, evidence on educational effectiveness and user impact is limited. Ethical concerns, including such as bias, model interpretability, and fairness, were commonly reported. While machine learning offers potential for enhancing post-diploma guidance through personalized recommendations and predictive insights, challenges remain inopacity, and limited explainability are acknowledged but not consistently addressed. The findings point to the need for more rigorous empirical validation and responsible implementation. Addressing these limitations is essential to ensuring the equitable and effective use of machine learning-driven career guidanceresearch, greater methodological transparency, and the integration of educational perspectives to ensure that ML-based systems for career guidance are used responsibly and with clear added value.