• 学术搜索
  • 科研智能体
    • Research Labs
    • AI 阅读
    • AI 文库
    • 深度研究
    • 学者亮点
  • 学术资源
    • AI2000
    • 期刊/会议
    • 学者库
    • 学术API
      MCP&Skill
    • 溯源树
    • 数据集
  • 知识沉淀
    • 学术空间
订阅小程序
旧版功能
aminer vip
开通会员低至0.73元/天
一次搞定AI科研
立即登录
  • English
  • 联系方式
    ACM Transactions on Recommender Systems

    ACM Transactions on Recommender Systems

    JournalISSN 2770-6699

    年发文量

    研究主题

    论文(242)

    排序
    1Fair and Online Itinerary Recommendation for Multiple Visitors in Crowded Destinations
    Jesse Nagel,Joris Slootweg, Caroline Jagtenberg,Sandjai Bhulai,Rob van der Mei

    Coordinating tourist trips in congested destinations allows for higher user satisfaction and reduced waiting times. A new online problem is introduced to allow the recommendation of tourist-trip itineraries when only limited information is available on arriving tourists. We propose a recommender system for generating congestion-aware itineraries in real time for a large number of tourists. The goal of the algorithm is to suggest high quality itinerary recommendations which lead to a fair allocation of tourists to points of interest. We test this method using different underlying heuristics. Numerical experiments show that the proposed algorithm outperforms a First-Come-First-Served policy and performs reasonably close to offline methods. The online algorithm is able to solve realistically sized problem instances with competitive running times. We find that for realistically sized instances, methods based on an iterated local search heuristic result in flatter distributions of score over users than those based on a greedy randomized adaptive search procedure.

    2026
    引用
    AI阅读
    加入学术空间
    2On Sampling Reliability in Offline Recommender Evaluation
    Bruno Pereira,Alan Said,Rodrygo Santos

    When online testing is impractical or risky, offline evaluation remains the primary mechanism for benchmarking recommender systems. Yet it is inherently affected by two sources of distortion: exposure bias, arising from system-driven item visibility, and sampling bias, introduced when evaluation relies on a subset of logged items rather than the full catalog. Prior work has proposed techniques to mitigate sampling bias. However, these techniques are typically assessed on static logs. Such assessments do not establish whether the techniques support reliable model comparisons under varying exposure conditions or against true user preferences. In this paper, we systematically study how logging and sampling choices jointly affect the reliability of offline evaluation. Using a fully observed dataset as ground truth, we simulate diverse exposure regimes and evaluate common sampling strategies along four dimensions: resolution (model separability), fidelity (agreement with full evaluation), robustness (stability under exposure bias), and predictive power (alignment with ground truth). We further examine whether these criteria align or reveal trade-offs in sampler performance. To strengthen external validity, we replicate key analyses on two additional benchmark datasets with fixed missing-completely-at-random (MCAR) test exposure. Our findings characterize when sampling distorts evaluation outcomes and provide controlled empirical evidence regarding how sampling strategies behave under different exposure and sparsity conditions.

    2026
    引用
    AI阅读
    加入学术空间
    3Playlist Generation from Titles: Transformers Vs. Large Language Models
    Enzo Charolois-Pasqua,Elea Vellard,Youssra Rebboud,Pasquale Lisena,Raphael Troncy

    The title of a playlist often reflects an intended mood or theme, allowing creators to easily locate their content and enabling other users to discover music that matches specific situations and needs. This work presents a novel approach to playlist generation using language models to leverage the thematic coherence between a playlist title and its tracks. Our method consists of creating semantic clusters from text embeddings, followed by fine-tuning a transformer model on these thematic clusters. Playlists are then generated considering the cosine similarity scores between known and unknown titles and applying a ranking mechanism, for which we compare a voting system and an LLM-based ranker. Performance evaluation, combining quantitative and qualitative metrics, demonstrates that using the playlist title as a seed provides useful recommendations, even in a zero-shot scenario.

    2026
    引用
    AI阅读
    加入学术空间
    4A Study of Normative Diversity Metrics in News Recommendations
    David Vos,Jin Huang,Maarten de Rijke

    Recent work on beyond-accuracy metrics for news recommendation has proposed RADio, a framework that operationalizes the concept of normative diversity from the social sciences into rank-aware evaluation metrics. RADio measures normative diversity as the divergence between ranked distributions over normative attributes of news articles. Vrijenhoek et al. caution that RADio’s conclusions may depend strongly on the dataset, cultural context, and the representation pipeline used to define such normative attributes. In this reproducibility study, we empirically study how robust RADio is to such contextual and representation choices. We evaluate the robustness of RADio along three axes. First, we extend RADio’s original evaluation from the English-language MIND dataset to the Danish EBNeRD and Norwegian Adressa news datasets, to test whether its metric behavior generalizes across editorial and cultural contexts. Second, we systematically vary key components of the attribute representation pipeline to examine how specific implementation choices influence RADio’s metric values. Finally, we analyze how RADio relates to recommendation accuracy by applying simple post-hoc re-ranking and quantifying the trade-offs between accuracy and normative diversity. Our results show that RADio outcomes are sensitive to both editorial context and representation design. Across datasets and representation choices, we observe that RADio can rank models’ normative diversity differently, and in some settings, the metric either saturates or collapses, limiting its ability to discriminate between models. We analyse why this happens, and give directions on designing the attribute representation pipeline. Furthermore, across all settings we consider, we can substantially increase RADio scores via re-ranking while incurring only modest losses in standard recommendation accuracy. These findings provide empirical support and nuance for RADio’s originally hypothesized contextual dependence, and underscore the need for context-aware evaluation frameworks when operationalizing normative diversity in news recommendations. In particular, the strong dependence on the attribute representation pipeline suggests that normative evaluation should be designed in close collaboration with editorial stakeholders. At the same time, our experiments indicate that, given appropriate representation choices, simple re-ranking can make normative-diversity-aware news recommendations practically viable. We share our code to facilitate reproducibility of our findings: https://github.com/davidvos/normative-diversity-repro /

    2026
    引用
    AI阅读
    加入学术空间
    5Reading Emotional Preferences in Adolescents: Emotion-Aware and Scalable Personalization for Book Recommendation
    Claire Cate, Ella Griffioen, Anna Felix, Kendra Schoonmaker,Yiu-Kai Ng

    Reading engagement during adolescence is strongly associated with long-term academic and emotional outcomes, yet recent large-scale evidence shows a persistent decline in teenagers’ reading motivation and proficiency. Although personalized recommender systems offer a promising educational technology, most existing book recommenders rely on behavioral logs or demographic assumptions, which are often sparse, biased, and difficult to deploy in privacy-sensitive environments such as schools and libraries. We propose a personalized, emotion-aware book recommendation framework that models adolescents’ individualized preferences using lightweight and interpretable emotional representations derived from publicly available book descriptions. Leveraging the NRC Emotion Intensity Lexicon (NRC-EIL), we embed users and books in a shared affective space and construct personalized target emotion vectors from previously preferred books. We evaluate three ranking strategies: (i) similarity-based retrieval in the emotion space, (ii) a hybrid method that conditions large language model (LLM) ranking on structured emotion vectors, and (iii) an LLM-only baseline without explicit affective modeling. Experiments on the Book-Crossing dataset demonstrate that incorporating structured emotional signals consistently improves ranking effectiveness across P@K, MRR, and NDCG metrics, with statistically significant gains validated using Wilcoxon signed-rank tests. Results further show robustness under cold-start and sparse-interaction settings while maintaining interpretability and avoiding reliance on sensitive user attributes. These findings establish emotion-aware retrieval as a principled and scalable direction for advancing personalized recommendation and supporting sustained teenage reading engagement.

    2026
    引用
    AI阅读
    加入学术空间
    立即登录,查看全部 242 篇论文

    高被引作者

    作者引用发文
    Yingrong Qin11441
    Yu Zheng11441
    Yuhan Quan11441
    Yinfeng Li11441
    Xiangnan He11441
    Chen Gao11441
    Nian Li11441
    Yong Li11441
    Jianxin Chang11441
    Depeng Jin11441

    高产作者

    作者引用发文
    Dietmar Jannach589
    Yashar Deldjoo877
    Alan Said457
    Yongfeng Zhang1707
    Li Chen97
    Zhong Ming1056
    Eva Zangerle355
    Weike Pan995
    Maarten de Rijke1065
    Fernando Diaz555

    相关期刊

    Foundations and Trends in Information Retrieval
    SIGKDD Explorations
    SIGWEB Newsletter
    ACM IKDD Conference on Data Science
    ACM Workshop on Wireless Security
    Workshop on E-Business
    Conference on Recommender Systems
    International Conference on the Theory of Information Retrieval
    Italian Information Retrieval Workshop
    arXiv: Information Retrieval