Graph Contrastive Learning (GCL) methods for recommendation learn representations by propagating signals over user-item interaction graphs. However, modeling these graphs with homogeneous edges can lead to semantic-structural misalignment, where information is exchanged between structurally adjacent but semantically dissimilar items, adversely affecting retrieval quality. Existing solutions typically rely on auxiliary encoders or additional supervision, increasing model complexity and training cost. We propose ACE (Affective Contrastive Embeddings), a framework that improves representation alignment by incorporating affective semantics into contrastive learning. ACE encodes the affective dimensions of valence and arousal as a topological prior, encouraging consistency between learned embeddings and an affective semantic space distilled from large language models. To operationalize this alignment, we introduce a Semantically Weighted Noise Contrastive Estimation (SW-NCE) loss that modulates contrastive gradients according to users' affective preferences. Experiments on Amazon, Last.fm, and SiTunes demonstrate that ACE consistently improves top-K retrieval performance over 11 baselines while reducing computational overhead. These results indicate that affective geometric alignment is an effective and efficient mechanism for enhancing graph-based retrieval models.

Graph models that manipulate the frequency spectrum of user-item interactions to separate preference signals from noise often report significant improvements, but concerns about evaluation rigor and reproducibility persist. We conduct a reproducibility and replicability study that examines three major families: (i) spectral denoising methods, (ii) graph signal processing (GSP) models, and (iii) spectral propagation approaches. Reproducing published pipelines reveals a polarized landscape: while several works are fully reproducible, others rely on flawed metric implementations and incomplete hyperparameter disclosures. In particular, we observe systematic inflation of Recall in the spectral denoising methods due to an implementation error, and theoretically invalid ranking metrics in GSP models due to unordered prediction lists; conversely, the graph filtering models are consistently reproducible. Beyond reproduction, we establish a unified evaluation protocol on four datasets with consistent splits and hyperparameter optimization for all baselines, showing that strong classical methods (e.g., SLIM, Item-kNN) remain highly competitive and that no single spectral model dominates across domains. We further analyze robustness under varying data sparsity and assess beyond-accuracy properties, finding that spectral filtering often improves catalog exploration even when accuracy gains are marginal. Our code is available at https://github.com/sisinflab/Mind_the_Metric_SIGIR-26.

Visual compatibility modeling is central to modern fashion recommendation systems. A key task is complementary item retrieval , where the goal is to identify a garment that harmonizes with a reference item, such as retrieving a compatible bottom for a given top. Recent generative approaches synthesize candidate garments to guide retrieval, but they either treat generation as an auxiliary signal or rely on computationally demanding architectures, limiting their practicality in large-scale deployments. In this work, we introduce GeCo , a generative–compositional framework that couples image synthesis and retrieval within an effective, two-stage design. In the first stage, a conditional GAN generates visually coherent bottom templates from top images; in the second stage, the generated template and top form a composed visual query for compatibility-based retrieval. This decoupled design avoids the heavy optimization pipelines used in prior generative approaches, resulting in stable training and low computational cost, while allowing the generated images to guide compatibility modeling. Experiments on three benchmarks, including the new FashionTaobaoTB dataset released with this work, show that GeCo offers competitive retrieval accuracy with a low memory footprint. Human evaluations further indicate that its generated items are perceived as realistic and stylistically compatible, supporting its suitability for practical, resource-constrained fashion recommendation scenarios. • GeCo : generative fashion compatibility model and complementary item retrieval. • GeCo uses a two-stage GAN under realistic hardware limits. • Human studies find its garments realistic and stylistically coherent. • FashionTaobaoTB and benchmarking codebase have been released.

Multimodal recommender systems (RSs) represent items in the catalog through multimodal data (e.g., product images and descriptions) that, in some cases, might be noisy or (even worse) missing. In those scenarios, the common practice is to drop items with missing modalities and train the multimodal RSs on a subsample of the original dataset. To date, the problem of missing modalities in multimodal recommendation has still received limited attention in the literature, lacking a precise formalisation as done with missing information in traditional machine learning. In this work, we first provide a problem formalisation for missing modalities in multimodal recommendation. Second, by leveraging the user-item graph structure, we re-cast the problem of missing multimodal information as a problem of graph features interpolation on the item-item co-purchase graph. On this basis, we propose four training-free approaches that propagate the available multimodal features throughout the item-item graph to impute the missing features. Extensive experiments on popular multimodal recommendation datasets demonstrate that our solutions can be seamlessly plugged into any existing multimodal RS and benchmarking framework while still preserving (or even widen) the performance gap between multimodal and traditional RSs. Moreover, we show that our graph-based techniques can perform better than traditional imputations in machine learning under different missing modalities settings. Finally, we analyse (for the first time in multimodal RSs) how feature homophily calculated on the item-item graph can influence our graph-based imputations.

Multimodal Recommender Systems aim to improve recommendation accuracy by integrating heterogeneous content, such as images and textual metadata. While effective, it remains unclear whether their gains stem from true multimodal understanding or increased model complexity. This work investigates the role of multimodal item embeddings, emphasizing the semantic informativeness of the representations. Initial experiments reveal that embeddings from standard extractors (e.g., ResNet50, Sentence-Bert) enhance performance, but rely on modality-specific encoders and ad hoc fusion strategies that lack control over cross-modal alignment. To overcome these limitations, we leverage Large Vision-Language Models (LVLMs) to generate multimodal-by-design embeddings via structured prompts. This approach yields semantically aligned representations without requiring any fusion. Experiments across multiple settings show notable performance improvements. Furthermore, LVLMs embeddings offer a distinctive advantage: they can be decoded into structured textual descriptions, enabling direct assessment of their multimodal comprehension. When such descriptions are incorporated as side content into recommender systems, they improve recommendation performance, empirically validating the semantic alignment encoded in LVLMs outputs. Our study highlights the importance of semantically rich representations and positions LVLMs as a compelling foundation to build robust and meaningful multimodal representations in recommendation tasks.

Pre-trained language models (PLMs) are widely used to derive semantic representations from item metadata in recommendation and search.In sequential recommendation, PLMs enhance ID-based embeddings through textual metadata, while in product search, they align item characteristics with user intent.Recent studies suggest task and domain-specific fine-tuning are needed to improve representational power.This paper challenges this assumption for e-commerce applications, showing that Generalist Text Embedding Models (GTEs), pre-trained on large-scale corpora, can guarantee strong zero-shot performance without specialized adaptation.Our experiments on popular e-commerce benchmarks demonstrate that GTEs outperform traditional and fine-tuned models in both sequential recommendation and product search.We attribute this to a superior representational power, as they distribute features more evenly across the embedding space.Finally, we show that compressing embedding dimensions by focusing on the most informative directions (e.g., via PCA) effectively reduces noise and improves the performance of specialized models.To ensure reproducibility, we provide our repository at https://github.com/sisinflab/GTE-Zero- Shot-Recsys.

In the realm of music recommendation, sequential recommenders have shown promise in capturing the dynamic nature of music consumption. A key characteristic of this domain is repetitive listening, where users frequently replay familiar tracks. To capture these repetition patterns, recent research has introduced Personalised Popularity Scores (PPS), which quantify user-specific preferences based on historical frequency. While PPS enhances relevance in recommendation, it often reinforces already-known content, limiting the system’s ability to surface novel or serendipitous items—key elements for fostering long-term user engagement and satisfaction. To address this limitation, we build upon RecJPQ, a Transformer-based framework initially developed to improve scalability in large-item catalogues through sub-item decomposition. We repurpose RecJPQ’s sub-item architecture to model personalised popularity at a finer granularity. This allows us to capture shared repetition patterns across sub-embeddings—latent structures not accessible through item-level popularity alone. We propose a novel integration of sub-ID-level personalised popularity within the RecJPQ framework, enabling explicit control over the trade-off between accuracy and personalised novelty. Our sub-ID-level PPS method (sPPS) consistently outperforms item-level PPS by achieving significantly higher personalised novelty without compromising recommendation accuracy. Code and experiments are publicly available at https://github.com/sisinflab/Sub-id-Popularity.

Recommendation pipelines involve several stages that can critically affect performance and reproducibility. However, early pipeline stages remain under-standardized, limiting comparability and interoperability across studies. This tutorial addresses this gap by providing both theoretical insights and hands-on experience with tools and practices for standardized data processing in recommender systems. In the first part, we introduce DataRec, a Python library for reproducible and interoperable data management, and discuss data filtering, splitting, and topological analysis techniques. In the second part, we explore multimodal feature extraction in domains such as fashion, music, and movies, focusing on the challenges of meaningful multimodal integration. We introduce Ducho, a unified framework for extracting audio, visual, and textual features using modern backends, and demonstrate its integration with the evaluation framework Elliot. The tutorial targets researchers and practitioners with an interest in recommender systems, data preprocessing, and multimodal modeling. All materials, including slides, code, datasets, and recordings, will be openly available on a dedicated tutorial website: https://sites.google.com/view/dd4rec-tutorial/.

Conversational Information Access systems have experienced widespread diffusion thanks to the natural and effortless interactions they enable with the user. In particular, they represent an effective interaction interface for conversational search (CS) and conversational recommendation (CR) scenarios. Despite their commonalities, CR and CS systems are often devised, developed, and evaluated as isolated components. Integrating these two elements would allow for handling complex information access scenarios, such as exploring unfamiliar recommended product aspects, enabling richer dialogues, and improving user satisfaction. As of today, the scarce availability of integrated datasets - focused exclusively on either of the tasks - limits the possibilities for evaluating by-design integrated CS and CR systems. To address this gap, we propose CoSRec, the first dataset for joint Conversational Search and Recommendation (CSR) evaluation. The CoSRec test set includes 20 high-quality conversations, with human-made annotations for the quality of conversations, and manually crafted relevance judgments for products and documents. Additionally, we provide supplementary training data comprising partially annotated dialogues and raw conversations to support diverse learning paradigms. CoSRec is the first resource to model CR and CS tasks in a unified framework, enabling the training and evaluation of systems that must shift between answering queries and making suggestions dynamically.

Recommender systems have demonstrated a significant impact across diverse domains, yet ensuring the reproducibility of experimental findings remains a persistent challenge.A primary obstacle lies in the fragmented and often opaque data management strategies employed during the preprocessing stage, where decisions about dataset selection, filtering, and splitting can substantially influence outcomes.To address these limitations, we introduce DataRec, an open-source Python-based library specifically designed to unify and streamline data handling in recommender system research.By providing reproducible routines for dataset preparation, data versioning, and seamless integration with other frameworks, DataRec promotes methodological standardization, interoperability, and comparability across different experimental setups.Our design is informed by an in-depth review of 55 stateof-the-art recommendation studies, ensuring that DataRec adopts best practices while addressing common pitfalls in data management.Ultimately, our contribution facilitates fair benchmarking, enhances reproducibility, and fosters greater trust in experimental results within the broader recommender systems community.The DataRec library, documentation, and examples are freely available at https://github.com/sisinflab/DataRec.

Large Language Models (LLMs) have become increasingly central to recommendation scenarios due to their remarkable natural language understanding and generation capabilities. Although significant research has explored the use of LLMs for various recommendation tasks, little effort has been dedicated to verifying whether they have memorized public recommendation dataset as part of their training data. This is undesirable because memorization reduces the generalizability of research findings, as benchmarking on memorized datasets does not guarantee generalization to unseen datasets. Furthermore, memorization can amplify biases, for example, some popular items may be recommended more frequently than others.

This work takes a critical stance on previous studies concerning fairness evaluation in Large-Language Model (LLM)-based recommender systems, which have primarily assessed consumer fairness by comparing recommendation lists generated with and without sensitive user attributes. Such approaches implicitly treat discrepancies in recommended items as biases, overlooking whether these changes might stem from genuine personalization aligned with true preferences of users. Moreover, these earlier studies typically address single sensitive attributes in isolation, neglecting the complex interplay of intersectional identities. In response to these shortcomings, we introduce CFaiRLLM , an enhanced evaluation framework that not only incorporates true preference alignment but also rigorously examines intersectional fairness by considering overlapping sensitive attributes. Additionally, CFaiRLLM introduces diverse user profile sampling strategies— random , top-rated , and recency-focused —to better understand the impact of profile generation fed to LLMs in light of inherent token limitations in these systems. Given that fairness depends on accurately understanding users’ tastes and preferences, these strategies provide a more realistic assessment of fairness within RecLLMs. To validate the efficacy of CFaiRLLM, we conducted extensive experiments using MovieLens and LastFM datasets, applying various sampling strategies and sensitive attribute configurations. The evaluation metrics include both item similarity measures and true preference alignment considering both hit and ranking (Jaccard Similarity and PRAG), thereby conducting a multi-faceted analysis of recommendation fairness. The results demonstrated that true preference alignment offers a more personalized and fair assessment compared to similarity-based measures, revealing significant disparities when sensitive and intersectional attributes are incorporated. Notably, our study finds that intersectional attributes amplify fairness gaps more prominently, especially in less structured domains such as music recommendations in LastFM. These findings suggest that future fairness evaluations in RecLLMs should incorporate true preference alignment to ensure equitable and genuinely personalized recommendations.

Recommender systems (RSs) provide customers with a personalized navigation experience within the vast catalogs of products and services offered on popular online platforms. Despite the substantial success of traditional RSs, recommendation remains a highly challenging task, especially in specific scenarios and domains. For example, human affinity for items described through multimedia content (e.g., images, audio, and text), such as fashion products, movies, and music, is multi-faceted and primarily driven by their diverse characteristics. Therefore, by leveraging all available signals in such scenarios, multimodality enables us to tap into richer information sources and construct more refined user/item profiles for recommendations. Despite the growing number of multimodal techniques proposed for multimedia recommendation, the existing literature lacks a shared and universal schema for modeling and solving the recommendation problem through the lens of multimodality. Given the recent advances in multimodal deep learning for other tasks and scenarios where precise theoretical and applicative procedures exist, we also consider it imperative to formalize a general multimodal schema for multimedia recommendation. In this work, we first provide a comprehensive literature review of multimodal approaches for multimedia recommendation from the last eight years. Second, we outline the theoretical foundations of a multimodal pipeline for multimedia recommendation by identifying and formally organizing recurring solutions/patterns; at the same time, we demonstrate its rationale by conceptually applying it to selected state-of-the-art approaches in multimedia recommendation. Third, we conduct a benchmarking analysis of recent algorithms for multimedia recommendation within Elliot, a rigorous framework for evaluating recommender systems, where we re-implement such multimedia recommendation approaches. Finally, we highlight the significant unresolved challenges in multimodal deep learning for multimedia recommendation and suggest possible avenues for addressing them. The primary aim of this work is to provide guidelines for designing and implementing the next generation of multimodal approaches in multimedia recommendation.

Generally, items with missing modalities are dropped in multimodal recommendation. However, with this work, we question this procedure, highlighting that it would further damage the pipeline of any multimodal recommender system. First, we show that the lack of (some) modalities is, in fact, a widely-diffused phenomenon in multimodal recommendation. Second, we propose a pipeline that imputes missing multimodal features in recommendation by leveraging traditional imputation strategies in machine learning. Then, given the graph structure of the recommendation data, we also propose three more effective imputation solutions that leverage the item-item co-purchase graph and the multimodal similarities of co-interacted items. Our method can be plugged into any multimodal RSs in the literature working as an untrained pre-processing phase, showing (through extensive experiments) that any data pre-filtering is not only unnecessary but also harmful to the performance.

In the realm of music recommendation, sequential recommender systems have shown promise in capturing the dynamic nature of music consumption. Nevertheless, traditional Transformer-based models, such as SASRec and BERT4Rec, while effective, encounter challenges due to the unique characteristics of music listening habits. In fact, existing models struggle to create a coherent listening experience due to rapidly evolving preferences. Moreover, music consumption is characterized by a prevalence of repeated listening, i.e. users frequently return to their favourite tracks, an important signal that could be framed as individual or personalized popularity. This paper addresses these challenges by introducing a novel approach that incorporates personalized popularity information into sequential recommendation. By combining user-item popularity scores with model-generated scores, our method effectively balances the exploration of new music with the satisfaction of user preferences. Experimental results demonstrate that a Personalized Most Popular recommender, a method solely based on user-specific popularity, outperforms existing state-of-the-art models. Furthermore, augmenting Transformer-based models with personalized popularity awareness yields superior performance, showing improvements ranging from 25.2% to 69.8%. The code for this paper is available at https://github.com/sisinflab/personalized-popularity-awareness.

Recently, graph neural networks (GNNs)-based recommender systems have encountered great success in recommendation. As the number of GNNs approaches rises, some works have started questioning the theoretical and empirical reasons behind their superior performance. Nevertheless, this investigation still disregards that GNNs treat the recommendation data as a topological graph structure. Building on this assumption, in this work, we provide a novel evaluation perspective on GNNs-based recommendation, which investigates the impact of the graph topology on the recommendation performance. To this end, we select some (topological) properties of the recommendation data and three GNNs-based recommender systems (i.e., LightGCN, DGCF, and SVD-GCN). Then, starting from three popular recommendation datasets (i.e., Yelp2018, Gowalla, and Amazon-Book) we sample them to obtain 1,800 size-reduced datasets that still resemble the original ones but can encompass a wider range of topological structures. We use this procedure to build a large pool of samples for which data characteristics and recommendation performance of the selected GNNs models are measured. Through an explanatory framework, we find strong correspondences between graph topology and GNNs performance, offering a novel evaluation perspective on these models.

Information Retrieval (IR) and Recommender Systems (RSs) tasks are moving from computing a ranking of final results based on a single metric to multi-objective problems. Solving these problems leads to a set of Pareto-optimal solutions, known as Pareto frontier, in which no objective can be further improved without hurting the others. In principle, all the points on the Pareto frontier are potential candidates to represent the best model selected with respect to the combination of two, or more, metrics. To our knowledge, there are no well-recognized strategies to decide which point should be selected on the frontier in IR and RSs. In this paper, we propose a novel, post-hoc, theoretically-justified technique, named "Population Distance from Utopia" (PDU), to identify and select the one-best Pareto-optimal solution. PDU considers fine-grained utopia points, and measures how far each point is from its utopia point, allowing to select solutions tailored to user preferences, a novel feature we call "calibration". We compare PDU against state-of-the-art strategies through extensive experiments on tasks from both IR and RS, showing that PDU combined with calibration notably impacts the solution selection.

The success of graph neural network-based models (GNNs) has significantly advanced recommender systems by effectively modeling users and items as a bipartite, undirected graph. However, many original graph-based works often adopt results from baseline papers without verifying their validity for the specific configuration under analysis. Our work addresses this issue by focusing on the replicability of results. We present a code that successfully replicates results from six popular and recent graph recommendation models (NGCF, DGCF, LightGCN, SGL, UltraGCN, and GFCF) on three common benchmark datasets (Gowalla, Yelp 2018, and Amazon Book). Additionally, we compare these graph models with traditional collaborative filtering models that historically performed well in offline evaluations. Furthermore, we extend our study to two new datasets (Allrecipes and BookCrossing) that lack established setups in existing literature. As the performance on these datasets differs from the previous benchmarks, we analyze the impact of specific dataset characteristics on recommendation accuracy. By investigating the information flow from users’ neighborhoods, we aim to identify which models are influenced by intrinsic features in the dataset structure. The code to reproduce our experiments is available at: https://github.com/sisinflab/Graph-RSs-Reproducibility.

Knowledge graphs (KG) have been proven to be a powerful source of side information to enhance the performance of recommendation algorithms. Their graph-based structure paves the way for the adoption of graph-aware learning models such as Graph Neural Networks (GNNs). In this respect, state-of-the-art models achieve good performance and interpretability via user-level combinations of intents leading users to their choices. Unfortunately, such results often come from and end-to-end learnings that considers a combination of the whole set of features contained in the KG without any analysis of the user decisions. In this paper, we introduce KGTORe, a GNN-based model that exploits KG to learn latent representations for the semantic features, and consequently, interpret the user decisions as a personal distillation of the item feature representations. Differently from previous models, KGTORe does not need to process the whole KG at training time but relies on a selection of the most discriminative features for the users, thus resulting in improved performance and personalization. Experimental results on three well-known datasets show that KGTORe achieves remarkable accuracy performance and several ablation studies demonstrate the effectiveness of its components. The implementation of KGTORe is available at: https://github.com/sisinflab/KGTORe.

Providing effective suggestions is of predominant importance for successful Recommender Systems (RSs). Nonetheless, the need of accounting for additional multiple objectives has become prominent, from both the final users’ and the item providers’ points of view. This need has led to a new class of RSs, called Multi-Objective Recommender Systems (MORSs). These systems are designed to provide suggestions by considering multiple (conflicting) objectives simultaneously, such as diverse, novel, and fairness-aware recommendations. In this work, we reproduce a state-of-the-art study on MORSs that exploits a reinforcement learning agent to satisfy three objectives, i.e., accuracy, diversity, and novelty of recommendations. The selected study is one of the few MORSs where the source code and datasets are released to ensure the reproducibility of the proposed approach. Interestingly, we find that some challenges arise when replicating the results of the original work, due to the nature of multiple-objective problems. We also extend the evaluation of the approach to analyze the impact of improving user-centered objectives of recommendations (i.e., diversity and novelty) in terms of algorithmic bias. To this end, we take into consideration both popularity and category of the items. We discover some interesting trends in the recommendation performance according to different evaluation metrics. In addition, we see that the multi-objective reinforcement learning approach is responsible for increasing the bias disparity in the output of the recommendation algorithm for those items belonging to positively/negatively biased categories. We publicly release datasets and codes in the following GitHub repository: https://github.com/sisinflab/MORS_reproducibility.

Although beyond-accuracy metrics have gained attention in the last decade, the accuracy of recommendations is still considered the gold standard to evaluate Recommender Systems (RSs). This approach prioritizes the accuracy of recommendations, neglecting the quality of suggestions to enhance user needs, such as diversity and novelty, as well as trustworthiness regulations in RSs for user and provider fairness. As a result, single metrics determine the success of RSs, but this approach fails to consider other criteria simultaneously. A downside of this method is that the most accurate model configuration may not excel in addressing the remaining criteria. This study seeks to broaden RS evaluation by introducing a multi-objective evaluation that considers all model configurations simultaneously under several perspectives. To achieve this, several hyper-parameter configurations of an RS model are trained, and the Pareto-optimal ones are retrieved. The Quality Indicators (QI) of Pareto frontiers, which are gaining interest in Multi-Objective Optimization research, are adapted to RSs. QI enables evaluating the model’s performance by considering various configurations and giving the same importance to each metric. The experiments show that this multi-objective evaluation overturns the ranking of performance among RSs, paving the way to revisit the evaluation approaches of the RecSys research community. We release codes and datasets in the following GitHub repository: https://github.com/sisinflab/RecMOE.

While the integration of product images enhances the recommendation performance of visual-based recommender systems (VRSs), this can make the model vulnerable to adversaries that can produce noised images capable to alter the recommendation behavior. Recently, stronger and stronger adversarial attacks have emerged to raise awareness of these risks; however, effective defense methods are still an urgent open challenge. In this work, we propose "Adversarial Image Denoiser" (AiD), a novel defense method that cleans up the item images by malicious perturbations. In particular, we design a training strategy whose denoising objective is to minimize both the visual differences between clean and adversarial images and preserve the ranking performance in authentic settings. We perform experiments to evaluate the efficacy of AiD using three state-of-the-art adversarial attacks mounted against standard VRSs. Code and datasets at https://github.com/sisinflab/Denoise-to-protect-VRS.

Collaborative filtering models have undoubtedly dominated the scene of recommender systems in recent years. However, due to the little use of content information, they narrowly focus on accuracy, disregarding a higher degree of personalization. Meanwhile, knowledge graphs are arousing considerable interest in recommendation models thanks to their ability to enrich the system with content features that captures subtle user-item relations. Nevertheless, with many high-quality features, the models become more complex and challenging to train. We extend KGFlex [ 16 ], a hybrid model that analyzes historical data to understand the semantic features the user decisions depend on. KGFlex represents item features as embeddings, and it models user-item interactions as a factorized entropy-driven combination of the item attributes relevant to the user, thus reducing the complexity and raising the degree of personalization. The method does not neglect long tail items, reducing the popularity bias and ensuring a high level of fairness. The user-item prediction is mediated by the user’s personal views of the embeddings that grant a high degree of expressiveness. This extension analyzes different entropy measurement strategies, an enhanced user negative decision modeling, and assesses the fairness of KGFlex and the impact of its hyperparameters. KGFlex is available at https://split.to/kgflex .

The advent of pretrained language have renovated the ways of handling natural languages, improving the quality of systems that rely on them. BERT played a crucial role in revolutionizing the Natural Language Processing (NLP) area. However, the deep learning framework it implements lacks interpretability. Thus, recent research efforts aimed to explain what BERT learns from the text sources exploited to pre-train its linguistic model. In this paper, we analyze the latent vector space resulting from the BERT context-aware word embeddings. We focus on assessing whether regions of the BERT vector space hold an explicit meaning attributable to a Knowledge Graph (KG). First, we prove the existence of explicitly meaningful areas through the Link Prediction (LP) task. Then, we demonstrate these regions being linked to explicit ontology concepts of a KG by learning classification patterns. To the best of our knowledge, this is the first attempt at interpreting the BERT learned linguistic knowledge through a KG relying on its pretrained context-aware word embeddings.

Due to their economic and significant importance, fault detection tasks in intelligent electrical grids are vital. Although numerous smart grid (SG) applications, such as fault detection and load forecasting, have adopted data-driven approaches, the robustness and security of these data-driven algorithms have not been widely examined. One of the greatest obstacles in the research of the security of smart grids is the lack of publicly accessible datasets that permit testing the system's resilience against various types of assault. In this paper, we present IEEE13-AdvAttack, a large-scaled simulated dataset based on the IEEE-13 test node feeder suitable for supervised tasks under SG. The dataset includes both conventional and renewable energy resources. We examine the robustness of fault type classification and fault zone classification systems to adversarial attacks. Through the release of datasets, benchmarking, and assessment of smart grid failure prediction systems against adversarial assaults, we seek to encourage the implementation of machine-learned security models in the context of smart grids. The benchmarking data and code for fault prediction are made publicly available on https://bit.ly/3NT5jxG.

Over the years, model-based approaches have shown their effectiveness in computing recommendation lists in different domains and settings. By relying on the computation of latent factors, they can recommend items with a very high level of accuracy. Unfortunately, when moving to the latent space, even if the model embeds content-based information, we miss references to the actual semantics of the recommended item. It makes the interpretation of the recommendation process non-trivial. In this paper, we show how to initialize latent factors in Factorization Machines by using semantic features coming from knowledge graphs to train an interpretable model, which is, in turn, able to provide recommendations with a high level of accuracy. In the presented approach, semantic features are injected into the learning process to retain the original informativeness of the items available in the dataset. By relying on the information encoded in the original knowledge graph, we also propose two metrics to evaluate the semantic accuracy and robustness of knowledge-aware interpretability. An extensive experimental evaluation on six different datasets shows the effectiveness of the interpretable model in terms of both accuracy and diversity of recommendation results and interpretability robustness.

Recommender systems (RSs) employ user-item feedback, e.g., ratings, to match customers to personalized lists of products. Approaches to top-k recommendation mainly rely on Learning-To-Rank algorithms and, among them, the most widely adopted is Bayesian Personalized Ranking (BPR), which bases on a pair-wise optimization approach. Recently, BPR has been found vulnerable against adversarial perturbations of its model parameters. Adversarial Personalized Ranking (APR) mitigates this issue by robustifying BPR via an adversarial training procedure. The empirical improvements of APR's accuracy performance on BPR have led to its wide use in several recommender models. However, a key overlooked aspect has been the beyond-accuracy performance of APR, i.e., novelty, coverage, and amplification of popularity bias, considering that recent results suggest that BPR, the building block of APR, is sensitive to the intensification of biases and reduction of recommendation novelty. In this work, we model the learning characteristics of the BPR and APR optimization frameworks to give mathematical evidence that, when the feedback data have a tailed distribution, APR amplifies the popularity bias more than BPR due to an unbalanced number of received positive updates from short-head items. Using matrix factorization (MF), we empirically validate the theoretical results by performing preliminary experiments on two public datasets to compare BPR-MF and APR-MF performance on accuracy and beyond-accuracy metrics. The experimental results consistently show the degradation of novelty and coverage measures and a worrying amplification of bias.

Recommender systems (RSs) have widely grown thanks to the outstanding capability of providing users with accurate and tailored recommendations. Recently, public awareness and new regulations forced RS researchers and practitioners to study solutions to user privacy endangerment. This tutorial will guide the attendees through the possible threats and the solutions towards private RSs.

The paper introduces Visual-Elliot (V-Elliot), a reproducibility framework for Visual Recommendation systems (VRSs) based on Elliot. framework provides the widest set of VRSs compared to other recommendation frameworks in the literature (i.e., 6 state-of-the-art models which have been commonly employed as baselines in recent works). The framework pipeline spans from the dataset preprocessing and item visual features loading to easily train and test complex combinations of visual models and evaluation settings. V-Elliot provides an extended set of features to ease the design, testing, and integration of novel VRSs into V-Elliot. The framework exploits of dataset filtering/splitting functions, 40 evaluation metrics, five hyper-parameter optimization methods, more than 50 recommendation algorithms, and two statistical hypothesis tests. The files of this demonstration are available at: github.com/sisinflab/elliot.

In the last few years, a renewed interest of the research community on conversational recommender systems (CRSs) is emerging. This is probably due to the great diffusion of Digital Assistants (DAs) such as Amazon Alexa, Siri, or Google Assistant that are revolutionizing the way users interact with machines. DAs allow users to execute a wide range of actions through an interaction mostly based on natural language messages. However, although DAs are able to complete tasks such as sending texts, making phone calls, or playing songs, they are still at an early stage on offering recommendation capabilities by using the conversational paradigm.

Collaborative filtering models based on matrix factorization and learned similarities using Artificial Neural Networks (ANNs) have gained significant attention in recent years. This is, in part, because ANNs have demonstrated very good results in a wide variety of recommendation tasks. However, the introduction of ANNs within the recommendation ecosystem has been recently questioned, raising several comparisons in terms of efficiency and effectiveness. One aspect most of these comparisons have in common is their focus on accuracy, neglecting other evaluation dimensions important for the recommendation, such as novelty, diversity, or accounting for biases. In this work, we replicate experiments from three different papers that compare Neural Collaborative Filtering (NCF) and Matrix Factorization (MF), to extend the analysis to other evaluation dimensions. First, our contribution shows that the experiments under analysis are entirely reproducible, and we extend the study including other accuracy metrics and two statistical hypothesis tests. Second, we investigated the Diversity and Novelty of the recommendations, showing that MF provides a better accuracy also on the long tail, although NCF provides a better item coverage and more diversified recommendation lists. Lastly, we discuss the bias effect generated by the tested methods. They show a relatively small bias, but other recommendation baselines, with competitive accuracy performance, consistently show to be less affected by this issue. This is the first work, to the best of our knowledge, where several complementary evaluation dimensions have been explored for an array of state-of-the-art algorithms covering recent adaptations of ANNs and MF. Hence, we aim to show the potential these techniques may have on beyond-accuracy evaluation while analyzing the effect on reproducibility these complementary dimensions may spark. The code to reproduce the experiments is publicly available on GitHub at https://tny.sh/Reenvisioning.

Recently, recommendation systems have been proven to be susceptible to malicious perturbations of the model weights. To overcome this vulnerability, Adversarial Regularization emerged as one of the most effective solutions. Interestingly, the technique not only robustifies the model, but also significantly increases its accuracy. To date, unfortunately, the effect of Adversarial Regularization beyond-accuracy evaluation dimensions is unknown. This paper sheds light on these aspects and investigates how Adversarial Regularization impacts the amplification of popularity bias, and the deterioration of novelty and coverage of the recommendation list. The results highlight that, with imbalanced data distribution, Adversarial Regularization amplifies the popularity bias. Moreover, the empirical validation on five datasets confirms that it degrades the diversity and novelty of the generated recommendation. Code and data are available at https://github.com/sisinflab/The-Idiosyncratic-Effects-of-Adversarial-Training.

Deep Learning and factorization-based collaborative filtering recommendation models have undoubtedly dominated the scene of recommender systems in recent years. However, despite their outstanding performance, these methods require a training time proportional to the size of the embeddings and it further increases when also side information is considered for the computation of the recommendation list. In fact, in these cases we have that with a large number of high-quality features, the resulting models are more complex and difficult to train. This paper addresses this problem by presenting KGFlex: a sparse factorization approach that grants an even greater degree of expressiveness. To achieve this result, KGFlex analyzes the historical data to understand the dimensions the user decisions depend on (e.g., movie direction, musical genre, nationality of book writer). KGFlex represents each item feature as an embedding and it models user-item interactions as a factorized entropy-driven combination of the item attributes relevant to the user. KGFlex facilitates the training process by letting users update only those relevant features on which they base their decisions. In other words, the user-item prediction is mediated by the user’s personal view that considers only relevant features. An extensive experimental evaluation shows the approach’s effectiveness, considering the recommendation results’ accuracy, diversity, and induced bias. The public implementation of KGFlex is available at https://split.to/kgflex.

Recommender Systems have shown to be an effective way to alleviate the over-choice problem and provide accurate and tailored recommendations. However, the impressive number of proposed recommendation algorithms, splitting strategies, evaluation protocols, metrics, and tasks, has made rigorous experimental evaluation particularly challenging. Puzzled and frustrated by the continuous recreation of appropriate evaluation benchmarks, experimental pipelines, hyperparameter optimization, and evaluation procedures, we have developed an exhaustive framework to address such needs. Elliot is a comprehensive recommendation framework that aims to run and reproduce an entire experimental pipeline by processing a simple configuration file. The framework loads, filters, and splits the data considering a vast set of strategies (13 splitting methods and 8 filtering approaches, from temporal training-test splitting to nested K-folds Cross-Validation). Elliot(https://github.com/sisinflab/elliot) optimizes hyperparameters (51 strategies) for several recommendation algorithms (50), selects the best models, compares them with the baselines providing intra-model statistics, computes metrics (36) spanning from accuracy to beyond-accuracy, bias, and fairness, and conducts statistical analysis (Wilcoxon and Paired t-test).

Visual-based recommender systems (VRSs) enhance recommendation performance by integrating users' feedback with the visual features of items' images. Recently, human-imperceptible image perturbations, defined adversarial samples, have been shown capable of altering the VRSs performance, for example, by pushing (promoting) or nuking (demoting) specific categories of products. One of the most effective adversarial defense methods is adversarial training (AT), which enhances the robustness of the model by incorporating adversarial samples into the training process and minimizing an adversarial risk. The AT effectiveness has been verified on defending DNNs in supervised learning tasks such as image classification. However, the extent to which AT can protect deep VRSs, against adversarial perturbation of images remains mostly under-investigated. This work focuses on the defensive side of VRSs and provides general insights that could be further exploited to broaden the frontier in the field. First, we introduce a suite of adversarial attacks against DNNs on top of VRSs, and defense strategies to counteract them. Next, we present an evaluation framework, named Visual Adversarial Recommender (VAR), to empirically investigate the performance of defended or undefended DNNs in various visually-aware item recommendation tasks. The results of large-scale experiments indicate alarming risks in protecting a VRS through the DNN robustification. Source code and data are available at https://github.com/sisinflab/Visual-Adversarial-Recommendation.

The impact of data characteristics on the performance of classical recommender systems has been recently investigated and produced fruitful results about the relationship they have with recommendation accuracy. This work provides a systematic study on the impact of broadly chosen data characteristics (DCs) of recommender systems. This is applied to the accuracy and fairness of several variations of CF recommendation models. We focus on a suite of DCs that capture properties about the structure of the user–item interaction matrix, the rating frequency, item properties, or the distribution of rating values. Experimental validation of the proposed system involved large-scale experiments by performing 23,400 recommendation simulations on three real-world datasets in the movie (ML-100K and ML-1M) and book domains (BookCrossing). The validation results show that the investigated DCs in some cases can have up to 90% of explanatory power – on several variations of classical CF algorithms –, while they can explain – in the best case – about 40% of fairness results (measured according to user gender and age sensitive attributes). Therefore, this work evidences that it is more difficult to explain variations in performance when dealing with fairness dimension than accuracy.

Adversarial Machine Learning (AML) has initially emerged as the field of study that investigates security issues of conventional and modern machine learning (ML) models. The objective of this tutorial is to present a comprehensive overview on the application of AML techniques for recommendation in a two-fold categorization: (i) AML for the attack/defense purposes, and (ii) AML to build GAN-based recommender models. A theoretical presentation on the topics is paired with two corresponding hands-on sessions to show the efficacy of AML application and push up novel ideas and advances in recommendation tasks. The tutorial is divided into four parts. We start by introducing a summary on state-of-the-art recommender models, including deep learning ones, and we define the fundamentals of AML. Then, we present the Adversarial Recommendation Framework, to represent attack/defense strategies on RSs, and the GAN-based Recommendation Framework, which is at the basis of novel adversarial-based generative recommenders. The presentation of each framework is followed by a practical session. Finally, we conclude with open challenges and possible future works for both applications.

Shilling attacks against collaborative filtering (CF) models are characterized by several fake user profiles mounted on the system by an adversarial party to harvest recommendation outcomes toward a malicious desire. The vulnerability of CF models is directly tied with their reliance on the underlying interaction data ---like user-item rating matrix (URM) --- to train their models and their inherent inability to distinguish genuine profiles from non-genuine ones. The majority of works conducted so far for analyzing shilling attacks mainly focused on properties such as confronted recommendation models, recommendation outputs, and even users under attack. The under-researched element has been the impact of data characteristics on the effectiveness of shilling attacks on CF models.

Recommender systems (RS) are an integral part of many online services aiming to provide an enhanced user-oriented experience. Machine learning (ML) models are nowadays broadly adopted in modern state-of-the-art approaches to recommendation, which are typically trained to maximize a user-centred utility (e.g., user satisfaction) or a business-oriented one (e.g., profitability or sales increase). They work under the main assumption that users' historical feedback can serve as proper ground-truth for model training and evaluation. However, driven by the success in the ML community, recent advances show that state-of-the-art recommendation approaches such as matrix factorization (MF) models or the ones based on deep neural networks can be vulnerable to adversarial perturbations applied on the input data. These adversarial samples can impede the ability for training high-quality MF models and can put the driven success of these approaches at high risk.

Over the last years, we have been witnessing the advent of more and more precise and powerful recommendation algorithms and techniques able to effectively assess users' tastes and predict information that would probably be of interest for them. Most of these approaches rely on the collaborative paradigm (often exploiting machine learning techniques) and do not take into account the huge amount of knowledge, both structured and non-structured ones, describing the domain of interest of the recommendation engine. Although very effective in in predicting relevant items, collaborative approaches miss some very interesting features that go beyond the accuracy of results and move into the direction of providing novel and diverse results as well as generating an explanation for the recommended items or support interactive and conversational recommendation processes.

Hyper-parameters tuning is a crucial task to make a model perform at its best. However, despite the well-established methodologies, some aspects of the tuning remain unexplored. As an example, it may affect not just accuracy but also novelty as well as it may depend on the adopted dataset. Moreover, sometimes it could be sufficient to concentrate on a single parameter only (or a few of them) instead of their overall set. In this paper we report on our investigation on hyper-parameters tuning by performing an extensive 10-Folds Cross-Validation on MovieLens and Amazon Movies for three well-known baselines: User-kNN, Item-kNN, BPR-MF. We adopted a grid search strategy considering approximately 15 values for each parameter, and we then evaluated each combination of parameters in terms of accuracy and novelty. We investigated the discriminative power of nDCG, Precision, Recall, MRR, EFD, EPC, and, finally, we analyzed the role of parameters on model evaluation for Cross-Validation.

More and more precise and powerful recommendation algorithms and techniques have been proposed over the last years able to effectively assess users' tastes and predict information that would probably be of interest for them. Most of these approaches rely on the collaborative paradigm (often exploiting machine learning techniques) and do not take into account the huge amount of knowledge, both structured and non-structured ones, describing the domain of interest for the recommendation engine. The aim of knowledge-aware and conversational recommender systems is to go beyond the traditional accuracy goal and to start a new generation of algorithms and interactive approaches which exploit the knowledge encoded in ontological and logic-based knowledge bases, knowledge graphs as well as the semantics emerging from the analysis and exploitation of semi-structured textual sources.

The Web has moved, slowly but steadily, from a collection of documents towards a collection of structured data. Knowledge graphs have then emerged as a way of representing the knowledge encoded in such data as well as a tool to reason on them in order to extract new and implicit information. Knowledge graphs are currently used, for example, to explain search results, to explore knowledge spaces, to semantically enrich textual documents, or to feed knowledge-intensive applications such as recommender systems. In this work, we describe how to create and exploit a knowledge graph to supply a hybrid recommendation engine with information that builds on top of a collections of documents describing musical and sound items. Tags and textual descriptions are exploited to extract and link entities to external graphs such as WordNet and DBpedia, which are in turn used to semantically enrich the initial data. By means of the knowledge graph we build, recommendations are computed using a feature combination hybrid approach. Two explicit graph feature mappings are formulated to obtain meaningful item feature representations able to catch the knowledge embedded in the graph. Those content features are further combined with additional collaborative information deriving from implicit user feedback. An extensive evaluation on historical data is performed over two different datasets: a dataset of sounds composed of tags, textual descriptions, and user’s download information gathered from Freesound.org and a dataset of songs that mixes song textual descriptions with tags and user’s listening habits extracted from Songfacts.com and Last.fm, respectively. Results show significant improvements with respect to state-of-the-art collaborative algorithms in both datasets. In addition, we show how the semantic expansion of the initial descriptions helps in achieving much better recommendation quality in terms of aggregated diversity and novelty.

Computing useful recommendations for cold-start users is a major challenge in the design of recommender systems, and additional data is often required to compensate the scarcity of user feedback. In this paper we address such problem in a target domain by exploiting user preferences from a related auxiliary domain. Following a rigorous methodology for cold-start, we evaluate a number of recommendation methods on a dataset with positive-only feedback in the movie and music domains, both in single and cross-domain scenarios. Comparing the methods in terms of item ranking accuracy, diversity and catalog coverage, we show that cross-domain preference data is useful to provide more accurate suggestions when user feedback in the target domain is scarce or not available at all, and may lead to more diverse recommendations depending on the target domain. Moreover, evaluating the impact of the user profile size and diversity in the source domain, we show that, in general, the quality of target recommendations increases with the size of the profile, but may deteriorate with too diverse profiles.

In most real-world scenarios, the ultimate goal of recommender system applications is to suggest a short ranked list of items, namely top- N recommendations, that will appeal to the end user. Often, the problem of computing top- N recommendations is mainly tackled with a two-step approach. The system focuses first on predicting the unknown ratings, which are eventually used to generate a ranked recommendation list. Actually, the top- N recommendation task can be directly seen as a ranking problem where the main goal is not to accurately predict ratings but to directly find the best-ranked list of items to recommend. In this article we present SPrank, a novel hybrid recommendation algorithm able to compute top- N recommendations exploiting freely available knowledge in the Web of Data. In particular, we employ DBpedia, a well-known encyclopedic knowledge base in the Linked Open Data cloud, to extract semantic path-based features and to eventually compute top- N recommendations in a learning-to-rank fashion. Experiments with three datasets related to different domains (books, music, and movies) prove the effectiveness of our approach compared to state-of-the-art recommendation algorithms.

Providing very accurate recommendations to end users has been nowadays recognized to be just one of the main tasks a recommender systems must be able to perform. While predicting relevant suggestions, attention needs to be paid to their diversification in order to avoid monotony in recommendation. In this paper we focus on modeling users' inclination toward selecting diverse items, where diversity is computed by means of content-based item attributes. We then exploit such modeling to present a novel approach to re-rank the list of Top-N items predicted by a recommendation algorithm, in order to foster diversity in the final ranking. Experimental evaluation proves the effectiveness of the proposed approach.

The primary goal of the workshop is to showcase cutting edge research on the intersection of Recommender Systems and Semantic Technologies, by taking the best of the two worlds. This combination may provide the RecSys community with important scenarios where the potential of Semantic Technologies can be effectively exploited into systems performing complex tasks, such as recommendation engines processing Big Data.

The advent of the Linked Open Data (LOD) initiative gave birth to a variety of open knowledge bases freely accessible on the Web. They provide a valuable source of information that can improve conventional recommender systems, if properly exploited. In this paper we present SPrank, a novel hybrid recommendation algorithm able to compute top-N item recommendations from implicit feedback exploiting the information available in the so called Web of Data. We leverage DBpedia, a well-known knowledge base in the LOD compass, to extract semantic path-based features and to eventually compute recommendations using a learning to rank algorithm. Experiments with datasets on two different domains show that the proposed approach outperforms in terms of prediction accuracy several state-of-the-art top-N recommendation algorithms for implicit feedback in situations affected by different degrees of data sparsity.

The availability of a huge amount of interconnected data in the so called Web of Data (WoD) paves the way to a new generation of applications able to exploit the information encoded in it. In this paper we present a model-based recommender system leveraging the datasets publicly available in the Linked Open Data (LOD) cloud as DBpedia and LinkedMDB. The proposed approach adapts support vector machine (SVM) to deal with RDF triples. We tested our system and showed its effectiveness by a comparison with different recommender systems techniques -- both content-based and collaborative filtering ones.

One of the main problems in online advertising is to display ads which are relevant and appropriate w.r.t. what the user is looking for. Often search engines fail to reach this goal as they do not consider semantics attached to keywords. In this paper we propose a system that tackles the problem by two different angles: help (i) advertisers to create more efficient ads campaigns and (ii) ads providers to properly match ads content to keywords in search engines. We exploit semantic relations stored in the DBpedia dataset and use an hybrid ranking system to rank keywords and to expand queries formulated by the user. Inputs of our ranking system are (i) the DBpedia dataset; (ii) external information sources such as classical search engine results and social tagging systems. We compare our approach with other RDF similarity measures, proving the validity of our algorithm with an extensive evaluation involving real users.

The paper focuses on an implementation of concept abduction with an Object-oriented Database Management System (OODBMS). OWL-DL Knowledge Bases have been translated to an OO version to enable standard and non-standard inference services as queries over a DB suitable for handheld devices. The framework has been implemented and tested: early experiments are reported.

In a generic semantic-based matchmaking process, given a request, it is desirable to obtain a ranked list of compatible services/resources/profiles in order of relevance. Furthermore, a match explanation can provide useful information to modify or refine the original request in a principled way. Though the feasibility of such an approach has been proved with fixed reasoning engines, it is a challenging subject to perform inference tasks on handheld devices. Here we propose abduction and contraction algorithms in Description Logics specifically devised for applications in mobile environments. A simple interaction paradigm based on Bluetooth protocol stack has also been implemented and tested in a mobile dating case study.

We propose a novel object discovery framework integrating the application layer of Bluetooth and RFID standards. The approach is motivated and illustrated in an innovative u-commerce setting. Given a request, it allows an advanced discovery process, exploiting semantically annotated descriptions of goods available in the u-marketplace. The RFID data exchange protocol and the Bluetooth service discovery protocol have been modified and enhanced to enable support for such semantic annotation of products. Modifications to the standards have been conceived to be backward compatible, thus allowing the smooth coexistence of the legacy discovery and/or identification features. Also noteworthy is the introduction of a dedicated compression tool to reduce storage/transmission problems due to the verbosity of XML-based semantic languages.

We propose a collaborative environment for semantic-enabled mobile devices (e.g. PDAs, cell phones, laptops) in peer to peer scenarios. Within the environment, resource discovery is performed exploiting technologies and techniques for knowledge representation developed for the semantic Web, which have been adapted to cope with the highly flexible structure of ad-hoc networks in ubiquitous computing. The approach exploits the standard Bluetooth stack, using the original UUID payload, to carry semantically annotated data. The environment is motivated and presented in a museum case study.

More and more resources are becoming available on the Web, and there is a growing need for infrastructures that, based on advertised descriptions, are able to semantically match demands with supplies.We formalize general properties a matchmaker should have, then we present a matchmaking facilitator, compliant with desired properties.The system embeds a NeoClassic reasoner, whose structural subsumption algorithm has been modified to allow match categorization into potential and partial, and ranking of matches within categories. Experiments carried out show the good correspondence between users and system rankings.