Adaptive Alleviation for Popularity Bias in Recommender Systems with Knowledge Graph

Recommender Systems (RS) tend to recommend more popular items instead of the relevant long-tail items. Mitigating such popularity bias is crucial to ensure that less popular but relevant items are part of the recommendation list shown to the user. In this work, we study the phenomenon of popularity bias in session-based RS (SRS) obtained via deep learning (DL) models. We observe that DL models trained on the historical user-item interactions in session logs (having long-tailed item-click distributions) tend to amplify popularity bias. To understand the source of this bias amplification, we consider potential sources of bias at two distinct stages in the modeling process: i. the data-generation stage (user-item interactions captured as session logs), ii. the DL model training stage. We highlight that the popularity of an item has a causal effect on i. user-item interactions via conformity bias, as well as ii. item ranking from DL models via biased training process due to class (target item) imbalance. While most existing approaches in literature address only one of these effects, we consider a comprehensive causal inference framework that identifies and mitigates the effects at both stages. Through extensive empirical evaluation on simulated and real-world datasets, we show that our approach improves upon several strong baselines from literature for popularity bias and long-tailed classification. Ablation studies show the advantage of our comprehensive causal analysis to identify and handle bias in data generation as well as training stages.

Popularity bias significantly limits the effectiveness of recommender systems by disproportionately favoring popular items and reducing exposure to diverse, less-known content. This bias negatively impacts personalization and marginalizes niche users and item providers. To address this challenge, calibrated recommendation methods have gained attention, notably the Calibrated Popularity (CP) approach, due to its simplicity, effectiveness, and model-agnostic nature. Originally, CP employs Jensen–Shannon divergence (JSD) to align the popularity distribution of recommended items with users’ historical interaction patterns. However, the choice of divergence measure substantially impacts calibration effectiveness and recommendation diversity. In this study, we systematically explore several alternative divergence measures, including Chi-Square, Wasserstein, Kullback–Leibler, Hellinger, Total Variation, Bhattacharyya, Cosine, and Renyi divergences, within the CP framework. Additionally, we propose a novel divergence-independent evaluation metric, namely Overall Similarity Error, enabling consistent assessment of calibration quality across divergence measures. Experimental results on two benchmark datasets using two collaborative filtering algorithms highlighted critical insights. More aggressive divergences, particularly Chi-Square, significantly enhanced calibration quality, reduced popularity bias, and increased recommendation diversity, albeit with a modest reduction in accuracy. In contrast, smoother divergences, such as JSD, maintained higher accuracy but provided limited improvements in reducing popularity bias. Also, the performed group-based analysis categorizing users into mainstream, balanced, and niche segments based on their historical popularity preferences revealed distinct patterns: balanced users typically achieved higher accuracy due to their evenly distributed preferences; mainstream users showed superior calibration results benefiting from robust signals of popular items; niche users obtained more diverse and personalized recommendations, clearly benefiting from aggressive divergence measures. These results underscore the complexity of addressing popularity bias and highlight the importance of adopting adaptive, user-aware calibration strategies to effectively balance accuracy, diversity, and fairness in recommender systems.

Knowledge graphs have emerged as a transformative tool in enhancing personalized recommendation systems. By integrating diverse datasets into a structured semantic network, knowledge graphs offer a holistic view of relationships and entities that can significantly improve the relevance and accuracy of recommendations. Unlike traditional recommendation algorithms that rely primarily on user behaviour and item similarity, knowledge graphs leverage contextual information and complex interconnections among entities to deliver more nuanced and context-aware suggestions. This abstract explores the pivotal role of knowledge graphs in advancing personalized recommendation systems, focusing on their ability to capture intricate relationships between users, items, and attributes. By mapping out these relationships, knowledge graphs facilitate a deeper understanding of user preferences and item characteristics, enabling the generation of more tailored and precise recommendations. Additionally, the incorporation of external knowledge sources into the graph can further enrich the recommendation process, leading to enhanced user satisfaction and engagement. The paper reviews various methodologies for integrating knowledge graphs into recommendation systems, including graph-based algorithms and machine learning techniques. It also examines real-world applications and case studies where knowledge graphs have demonstrated substantial improvements in recommendation quality. Ultimately, the utilization of knowledge graphs represents a significant leap forward in personalizing user experiences, offering a promising avenue for future research and development in the field of recommendation systems.

Investigating and counteracting popularity bias in group recommendations

Social recommendation with Graph Neural Networks(GNNs) learns to represent cold users by fusing user-user social relations with user-item interactions, thereby alleviating the cold-start problem associated with recommender systems. Despite being well adapted to social relations and user-item interactions, these supervised models are still susceptible to popularity bias. Contrastive learning helps resolve this dilemma by identifying the properties that distinguish positive from negative samples. In its previous combinations with recommender systems, social relationships and cold-start cases in this context are not considered. Also, they primarily focus on collaborative features between users and items, leaving the similarity between items under-utilized. In this work, we propose socially-aware dual contrastive learning for cold-start recommendation, where cold users can be modeled in the same way as warm users. To take full advantage of social relations, we create dynamic node embeddings for each user by aggregating information from different neighbors according to each different query item, in the form of user-item pairs. We further design a dual-branch self-supervised contrastive objective to account for user-item collaborative features and item-item mutual information, respectively. On one hand, our framework eliminates popularity bias with proper negative sampling in contrastive learning, without extra ground-truth supervision. On the other hand, we extend previous contrastive learning methods to provide a solution to cold-start problem with social relations included. Extensive experiments on two real-world social recommendation datasets demonstrate its effectiveness.

Recommender systems are frequently evaluated using performance indexes based on variants and extensions of precision-like measures. As these measures are biased toward popular items, a list of recommendations simply must include a few popular items to perform well. To address the popularity bias challenge, new approaches for novelty and diversity evaluation have been proposed. On the one hand, novelty-based approaches model the quality of being new as apposed to that which is already known. Novelty approaches are commonly based on item views or user rates. On the other hand, diversity approaches model the quality of an item that is composed of different content elements. Diversity measures are commonly rooted in content-based features that characterize the diversity of the content of an item in terms of the presence/absence of a number of predefined nuggets of information. As item contents are also biased to popular contents (e.g., drama in movies or pop in music), diversity-based measures are also popularity biased. To alleviate the effect of popularity bias on diversity measures, we used an evaluation approach based on the degree of novelty of the elements that make up each item. We named this approach content novelty, as it mixes content and diversity approaches in a single and coherent evaluation framework. Experimental results show that our proposal is feasible and useful. Our findings demonstrate that the proposed measures yield consistent and interpretable results, producing insights that reduce the impact of popularity bias in the evaluation of recommender systems.

Whilepersonalizationincreases the utility of item recommendation, it also suffers from the issue of popularity bias. However, previous methods emphasize adopting supervised learning models to relieve popularity bias in the static recommendation, ignoring the dynamic transfer of user preference and amplification effects of the feedback loop in the recommender system (RS). In this paper, we focus on studying this issue in the interactive recommendation. We argue that diversification and novelty are both equally crucial for improving user satisfaction of IRS in the aforementioned setting. To achieve this goal, we propose aDiversity-Novelty-awareInteractiveRecommendation framework (DNaIR) that augments offline reinforcement learning (RL) to increase the exposure rate of long-tail items with high quality. Its main idea is first to aggregate the item similarity, popularity, and quality into the reward model to help the planning of RL policy. It then designs a diversity-aware stochastic action generator to achieve an efficient and lightweight DNaIR algorithm. Extensive experiments are conducted on the three real-world datasets and an authentic RL environment (Virtual-Taobao). The experiments show that our model can better and full use of the long-tail items to improve recommendation satisfaction, especially those low popularity items with high-quality ones, thus achieving state-of-the-art performance.

Knowledge Graph (KG), which commonly consists of fruitful connected facts about items, presents an unprecedented opportunity to alleviate the sparsity problem in recommender system. However, existing KG based recommendation methods mainly rely on handcrafted meta-path features or simple triple-level entity embedding, which cannot automatically capture entities' long-term relational dependencies for the recommendation. In this paper, a two-channel neural interaction method named Knowledge Graph enhanced Neural Collaborative Filtering with Residual Recurrent Network (KGNCF-RRN) is proposed, which leverages both long-term relational dependencies KG context and user-item interaction for recommendation. (1) For the KG context interaction channel, we propose a Residual Recurrent Network (RRN) to construct context-based path embedding, which incorporates residual learning into traditional recurrent neural networks (RNNs) to efficiently encode the long-term relational dependencies of KG. The self-attention network is then applied to the path embedding to capture the polysemy of various user interaction behaviours. (2) For the user-item interaction channel, the user and item embeddings are fed into a newly designed two-dimensional interaction map. (3) Finally, above the two-channel neural interaction matrix, we employ a convolutional neural network to learn complex correlations between user and item. Extensive experimental results on three benchmark datasets show that our proposed approach outperforms existing state-of-the-art approaches for knowledge graph based recommendation.

Multimedia recommender systems suggest media items, e.g., songs, (digital) books and movies, to users by utilizing concepts of traditional recommender systems such as collaborative filtering. In this paper, we investigate a potential issue of such collaborative-filtering based multimedia recommender systems, namely popularity bias that leads to the underrepresentation of unpopular items in the recommendation lists. Therefore, we study four multimedia datasets, i.e., Last.fm, MovieLens, BookCrossing and MyAnimeList, that we each split into three user groups differing in their inclination to popularity, i.e., LowPop, MedPop and HighPop. Using these user groups, we evaluate four collaborative filtering-based algorithms with respect to popularity bias on the item and the user level. Our findings are three-fold: firstly, we show that users with little interest into popular items tend to have large user profiles and thus, are important data sources for multimedia recommender systems. Secondly, we find that popular items are recommended more frequently than unpopular ones. Thirdly, we find that users with little interest into popular items receive significantly worse recommendations than users with medium or high interest into popularity.Keywordsmultimedia recommender systemscollaborative filteringpopularity biasalgorithmic fairness

With the advancement in recommendation techniques, focus is diverted from just making them more accurate to making them fairer and diverse, thus catering to the set of less-popular items (the long tail) that often get neglected due to inherent biases in recommender systems. Popularity bias has been recently acknowledged as a major bias of critical concern in the field of recommender systems. Although research on popularity bias has gained pace from the last couple of years, this field is believed to be still in its infancy. To advance research in this area, this paper thoroughly investigates current state of the art and could have a very positive impact on further research in popularity bias. Besides the mitigation techniques discussed in this paper, allied evaluation metrics that were used in measuring popularity bias have also been discussed.KeywordsPopularity biasAlgorithmic biasRecommender systems

Multitype view of knowledge contrastive learning for recommendation

Learning instrumental variable representation for debiasing in recommender systems.

Collaborative filtering and neural recommendation models have advanced substantially over the past decade, yet they frequently struggle to capture the fine-grained structure of user preferences and the rich, multi-relational dependencies among items. Knowledge graph-based methods provide a principled means of representing such structured information; however, systematic mechanisms for combining structured reasoning with heterogeneous recommendation signals remain relatively underexplored. In this work, we introduce KGERA (Knowledge Graph Enhanced Reasoning Architecture), a recommendation framework that augments a multi-relational knowledge graph with an explicit reasoning module to enhance ranking quality through test-time inference. We construct a unified knowledge graph encoding diverse item relationships, including genre co-membership, creator-level associations, and data-driven similarity links, and apply structured reasoning on this graph to compute contextualized recommendation scores. Unlike approaches that require expensive large language model inference, KGERA performs lightweight, interpretable reasoning exclusively at inference time using embedding-based computations, allowing the system to adapt dynamically to evolving user preference signals without retraining. The reasoning component is coupled with nine complementary recommendation strategies through a learned ensemble that integrates collaborative filtering (ItemKNN), graph neural networks (LightGCN), self-supervised graph learning (SGL, SimGCL), neural collaborative filtering (NCF), knowledge graph embeddings (TransE), content-based filtering (TextProxy), and popularity-based signals (PopRec) into a single predictive model. Experiments on MovieLens-1M demonstrate that KGERA significantly outperforms strong baselines, including recent self-supervised methods from SIGIR 2021-2022, achieving improvements of 52.83% in NDCG@10, 50.24% in Recall@10, and 54.62% in MRR@10 over the best baseline. Comprehensive validation through leave-one-out ablation studies, threshold sensitivity analysis, 5-fold cross-validation (CV = 4.84%), and stratification experiments across user activity levels (77.5%–85.6% improvement for low/medium activity users), item popularity tiers (114.2% improvement for medium-popularity items), and content genres demonstrates the robustness and generalizability of our approach. All improvements are statistically significant with $$p < 0.001$$ under paired t-tests. Collectively, these results indicate that explicit test-time reasoning over structured knowledge provides a computationally efficient and interpretable direction for advancing the design of next-generation recommender systems.

Recent studies have shown that the problems of cold processing and data sparsity in recommender systems can be well alleviated with the help of knowledge graph information. Meanwhile, since the information in recommender systems can be represented by graph data structure, graph neural network has great advantages in recommendatory system. Therefore, graph neural network can better mine the deep relationship between objects in the recommender systems which combines with information from knowledge graph. This paper reviews the application of graph neural network in knowledge graph aided recommendation systems. Specifically, this paper firstly sorts out the related knowledge of knowledge graph, recommendation system and graph neural network, including the cold processing and data sparsity, the structure of knowledge graph, and the basic modules of graph neural network. This paper also summarizes the methods such as KGAT and CKAN used in knowledge graph aided recommendation systems in recent years, and gives some discussion on the development of this field.

To address the sparsity and cold-start problem of collaborative filtering, researchers usually make use of side information, such as social networks or item attributes, to improve the performance of recommendation. In this article, we consider the knowledge graph (KG) as the source of side information. To address the limitations of existing embedding-based and path-based methods for KG-aware recommendation, we propose RippleNet , an end-to-end framework that naturally incorporates the KG into recommender systems. RippleNet has two versions: (1) The outward propagation version, which is analogous to the actual ripples on water, stimulates the propagation of user preferences over the set of knowledge entities by automatically and iteratively extending a user’s potential interests along links in the KG. The multiple “ripples” activated by a user’s historically clicked items are thus superposed to form the preference distribution of the user with respect to a candidate item. (2) The inward aggregation version aggregates and incorporates the neighborhood information biasedly when computing the representation of a given entity. The neighborhood can be extended to multiple hops away to model high-order proximity and capture users’ long-distance interests. In addition, we intuitively demonstrate how a KG assists with recommender systems in RippleNet, and we also find that RippleNet provides a new perspective of explainability for the recommended results in terms of the KG. Through extensive experiments on real-world datasets, we demonstrate that both versions of RippleNet achieve substantial gains in a variety of scenarios, including movie, book, and news recommendations, over several state-of-the-art baselines.