1 Introduction

Group recommender systems (Felfernig et al. 2024; Masthoff 2015) determine recommendations for groups. These systems often apply basic recommendation strategies and try to aggregate the results in such a way that recommendations become acceptable for all or most group members.

However, there are still plenty of open issues to be solved, especially due to the nature of group decision scenarios. In contrast with single-user recommender systems, recommender systems for groups must consider aspects such as group dynamics, emotions, personality, and heterogeneity of groups. These factors make recommender-based decision support a challenge with a multitude of related research questions also including psychological aspects, for example, “How to best take into account models of human decision- making in group-based recommendation scenarios?” (Tran et al. 2021).

Moreover, group settings are becoming prominent in a variety of scenarios, such as conversational systems, which could help exploiting the interaction between group members, and algorithmic fairness, where groups play a key role to assess possible forms of discrimination in the recommendation process. In addition to this, the relation of group recommendation with respect to recent regulations, such as GDPR, which require algorithms to be explainable, privacy preserving, and unbiased, opens a plethora of relevant issues to tackle in the group recommendation research (Ogunseyi et al. 2023).

This special issue brings together original research methods and applications related to group recommender systems. The remainder of this article is structured as follows: Sect. 2 summarizes the contributions included in this special issue, and Sect. 3 provides concluding remarks.

2 Content of this special issue

This special issue brings together six articles that explore various aspects of group recommender systems (GRS), ranging from consensus-based models and privacy dynamics, through explainable aggregation strategies and deep learning methods, to personality influences and genetic algorithms for group formation, and their applications across different domains.

In the article “An overview of consensus models for group decision-making and group recommender systems,” by Tran et al. (2023), the state of the art in consensus-based group recommender systems is reviewed, emphasizing the importance of consensus in group decision-making. The authors categorize various soft consensus models, which aim to achieve agreement among group members without requiring complete unanimity. The review includes 80 studies and highlights approaches integrating consensus models into recommender systems. The authors provide a classification of consensus methods, an analysis of their application in recommender systems, and an identification of challenges and future research directions.

In their study “How do people make decisions in disclosing personal information in tourism group recommendations in competitive versus cooperative conditions?,” Najafian et al. (2023) investigate the dynamics of personal information disclosure in group recommendation settings, emphasizing the impact of privacy concerns on decision-making. The study addresses the gap between individuals’ privacy attitudes and their actual disclosure behaviors by examining factors such as personality traits, preference alignment, and the nature of the group (e.g., loosely coupled heterogeneous groups). The authors reveal that these factors significantly influence general privacy perception, which affects trust in the group and perceived privacy risks. Privacy risk predicts disclosure behavior in competitive tasks but not in cooperative ones, providing insights for designing GRSs that balance privacy concerns with effective group decision-making.

Barile et al. (2024) in their study “Evaluating explainable social choice-based aggregation strategies for group recommendation,” examine the effectiveness of social choice-based aggregation strategies in group recommender systems (GRSs) and their impact on users’ perceptions of fairness, consensus, and satisfaction. Through two experiments, the study evaluates different strategies in diverse group scenarios, revealing significant differences in their effectiveness depending on group configurations such as uniform, divergent, coalitional, and minority. Findings indicate that strategies like Most Pleasure and Fairness perform variably across different group types, highlighting the importance of context in strategy selection. Moreover, adding explanations to these strategies did not enhance user perceptions of fairness or consensus.

In the article “Deep adversarial group recommendation with user feature space separation,” Sun et al. (2023) introduce a deep adversarial group recommendation method called DA-GR, designed to address data sparsity and complex relationship dynamics in group recommender systems. Traditional systems often fail to accurately model group preferences due to limited interaction data and the simplistic aggregation of individual preferences. DA-GR separates user preferences into private and shared subspaces, capturing distinct individual behaviors, both independently and within a group. By employing adversarial learning, DA-GR facilitates effective knowledge transfer between individual users and groups, enhancing recommendation quality. Experimental results on real-world datasets show that DA-GR significantly improves recommendation performance for both individuals and groups, effectively mitigating data sparsity challenges.

Alves et al., in their study “Group recommender systems for tourism: how does personality predict preferences for attractions, travel motivations, preferences and concerns?,” (Alves et al., 2023) explore the impact of personality on tourist preferences, motivations, and travel-related concerns in group recommender systems for tourism. Leveraging the Big Five personality dimensions (Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism), the study investigates how these traits influence various travel aspects. The research aims to improve GRSs by automatically modeling tourist profiles based on personality, thus addressing issues like the cold-start problem and conflicting group preferences. Extensive data collection and analysis revealed that personality significantly predicts preferences for tourist attractions and travel-related concerns, although only neuroticism and openness were linked to travel motivations. The study proposes three models correlating personality with travel preferences and a fourth model for subgroup creation to mitigate conflicting interests.

In the article “A novel group recommender system for domain-independent decision support customizing a grouping genetic algorithm,” by Krouska et al. (2023), a novel genetic algorithm-based group recommender system is presented, designed to improve group formation processes across various domains. Group formation, a critical task in settings such as education, healthcare, and industry, benefits from algorithmic approaches that optimize team performance and member satisfaction. Traditional methods like probabilistic algorithms, data mining, and multi-agent systems have been widely used, but genetic algorithms are particularly effective due to their ability to handle numerous variables and generate optimal solutions. The proposed system introduces innovative genetic operators and allows users to customize the algorithm settings to achieve the desired group formation outcomes. This flexibility ensures that the system can handle different types of inputs, whether grouping people, objects, or items, making it domain independent. The system also balances execution time and accuracy by adjusting the number of operators used.

3 Conclusions

Understanding the dynamics and challenges of group recommender systems is essential for their effective applications across various domains. The articles included in this special issue address a wide range of topics, from consensus models and privacy dynamics to explainable aggregation strategies, deep learning methods, and the influence of personality traits. These contributions not only enhance our current understanding of group recommender systems but also highlight innovative approaches to improving their performance and user satisfaction. We hope that the insights and methodologies presented in these articles will provide the community with a comprehensive view of the current state of research in group recommender systems and inspire further exploration and development in this rapidly evolving field.