Many news recommender systems (NRS) adopt a one-recommenderfor-all approach, overlooking that users engage with news in fundamentally different ways.In this work, we identify user clusters based on various engagement metrics that go beyond clicks by employing cluster analysis on two real-world datasets: EB-NeRD and Adressa.Next to that, we evaluate the performance of common recommender strategies: popularity, collaborative filtering (EASE and ItemKNN), and a content-based model across these user clusters, which exhibit varying reading behaviors and information needs.Our findings show that different recommender strategies are effective to varying degrees depending on the user cluster.This study contributes to NRS research by providing a grounded clustering of users derived from real-world datasets and emphasizes the importance of user-centered evaluations for understanding how NRS strategies serve audiences with varying levels of news engagement.

Previous research has used Large Language Models (LLMs) to develop personalized Conversational Recommender Systems (CRS) with text-based user interfaces (UIs). However, the potential of LLMs to generate interactive graphical elements that enhance user experience remains largely unexplored. To address this gap, we introduce "GenUI(ne) CRS," a novel framework designed to leverage LLMs for adaptive and interactive UIs. Our framework supports domain-specific graphical elements such as buttons and cards, in addition to text-based inputs. It also addresses the common LLM issue of outdated knowledge, known as the "knowledge cut-off," by implementing Retrieval-Augmented Generation (RAG). To illustrate its potential, we developed a prototype movie CRS. This work demonstrates the feasibility of LLM-powered interactive UIs and paves the way for future CRS research, including user experience validation, transparent explanations, and addressing LLM biases.

Recommender systems are increasingly used by news media organizations. Existing literature examines various aspects of news recommender systems (NRS) from a computational, user-centric, or normative perspective. Yet research advocates studying the complexities of real-world applications around NRS. Recently, a multi-stakeholder approach to NRS has been adopted, allowing to understand different stakeholder perspectives on NRS development and evaluation within the news organization. However, little research has been done on the different key performance indicators (KPIs) and metrics considered valuable by different stakeholders. Based on 11 interviews with professionals from two commercial news publishers, this paper demonstrates that stakeholders prioritize distinct KPIs and metrics related to the reach-engagement-conversion-retention funnel. The evaluation of NRS performance is often limited to short-term metrics like CTR, overlooking the multiplicity of stakeholders involved. Our findings reveal how different purposes, KPIs, and metrics are valued from the journalistic, commercial, and tech logic. In doing so, this paper contributes to the multi-stakeholder approach to NRS, advancing our understanding of the real-world complexity of NRS development and evaluation.

News media play an important role in democratic societies. Central to fulfilling this role is the premise that users should be exposed to diverse news. However, news recommender systems are gaining popularity on news websites, which has sparked concerns over filter bubbles. More specifically, editors, policy-makers and scholars are worried that these news recommender systems may expose users to less diverse content over time. To the best of our knowledge, this hypothesis has not been tested in a longitudinal observational study of real users that interact with a real news website. Such observational studies require the use of research methods that are robust and can account for the many covariates that may influence the diversity of recommendations at any given time. In this work, we propose an analysis model to study whether the variety of articles recommended to a user decreases over time in such an observational study design. Further, we present results from two case studies using aggregated and anonymized data that were collected by two western European news websites employing a collaborative filtering-based news recommender system to serve (personalized) recommendations to their users. Through these case studies we validate empirically that our modeling assumptions are sound and supported by the data, and that our model obtains more reliable and interpretable results than analysis methods used in prior empirical work on filter bubbles. Our case studies provide evidence of a small decrease in the topic variety of a user’s recommendations in the first weeks after they sign up, but no evidence of a decrease in political variety.

Abstract The contemporary city is increasingly being labeled as a smart city consisting of both physical and virtual spaces. This digital augmentation of urban life sets the scene for urban recommender systems to help citizens dealing with the abundance of digital information and corresponding choice overload, for example, by recommending the best place to have dinner based on your personal profile. There are, however, concerns that this kind of algorithmic filtering could lead to homogenization of urban experiences and a decline of social cohesion among citizens. To overcome this issue, scholars increasingly encourage the introduction of serendipity in all types of recommender systems. Nonetheless, it remains unclear how this can be achieved in practice. In this work, we study user evaluations of serendipity in urban recommender systems through a survey among 1,641 citizens. More specifically, we study which characteristics of recommended items contribute to serendipitous experiences and to what extent this increases user satisfaction and conversion. Our results align with findings in other application domains in the sense that there is a strong relation between the relevance and novelty of recommendations and the corresponding experienced serendipity. Moreover, serendipitous recommendations are found to increase the chance of users following up on these recommendations.