Quality Of Recommendations Research Articles

Design Method for Travel E-commerce Platform Based on HHO imparoved K-means Clustering Algorithm

Convenient and intelligent tourism product recommendation method, as the key technology of tourism E-commerce platform design, not only provides academic value to the research of tourism E-commerce platform, but also improves the efficiency of personalized recommendation of tourism products. In order to improve the quality of tourism recommendation, this paper proposes a tourism E-commerce platform design method based on HHO improved K-means clustering algorithm. Firstly, the Harris optimization algorithm is used to improve the K-means algorithm to construct a user-oriented tourism product recommendation strategy; then, combined with the XGBoost algorithm, an item-oriented tourism product recommendation strategy is proposed; secondly, the two strategies are mixed to construct a personalized tourism product recommendation model. Finally, the effectiveness of the proposed method is verified by simulation experiment analysis. The results show that the recommendation accuracy of the tourism E-commerce platform design method proposed in this paper reaches more than 90%, and the recommendation response time meets the real-time requirements, which can provide personalized tourism product recommendation for platform users and enhance the purchase of tourism products.

E-Commerce Recommendation System

In today's digitally driven marketplace, E-commerce Product Recommendation Systems (EPRS) play a pivotal role in enhancing user experience, facilitating decision-making, and boosting sales. This paper offers a comprehensive analysis of various methodologies, techniques, and advancements in EPRS. Beginning with an overview of the significance of personalized recommendations in driving user engagement and satisfaction, the paper delves into the underlying principles and algorithms employed in EPRS, including collaborative filtering, content-based filtering, and hybrid approaches. Furthermore, this research explores the challenges associated with EPRS, such as data sparsity, cold start problem, and scalability issues, along with the strategies and innovations proposed to address these challenges. It examines the role of machine learning, deep learning, and artificial intelligence in improving the accuracy and relevance of recommendations, thereby optimizing user experience and maximizing conversion rates. Moreover, the paper investigates the impact of contextual factors, such as user demographics, browsing history, and social interactions, on recommendation quality and effectiveness. It discusses the ethical considerations and privacy concerns surrounding data collection, user profiling, and algorithmic bias in EPRS implementation, emphasizing the need for transparency, fairness, and user consent. Additionally, this research evaluates the performance metrics and evaluation methodologies used to assess the effectiveness and efficiency of EPRS, including precision, recall, coverage, and serendipity. It highlights the importance of continuous evaluation and refinement of recommendation algorithms to adapt to evolving user preferences and market dynamics. In conclusion, this paper provides valuable insights into the state-of-the-art techniques, challenges, and future directions of E-commerce Product Recommendation Systems. By understanding the intricacies and advancements in this field, businesses can leverage EPRS to enhance customer satisfaction, foster brand loyalty, and drive sustainable growth in the competitive landscape of ecommerce. keyword Recommendation Algorithms,

Hyperbolic multichannel graph convolutional network for social recommendation

Social recommendations enhance the quality of recommendations by integrating social network information. Existing methods predominantly rely on pairwise relationships to uncover potential user preferences. However, they usually overlook the exploration of higher-order user relations. Moreover, because social relation graphs often exhibit scale-free graph structures, directly embedding them in Euclidean space will lead to significant distortion. To this end, we propose a novel graph neural network framework with hypergraph and hyperbolic embedding learning, namely HMGCN. Specifically, we first construct hypergraphs over user-item interactions and social networks, and then perform graph convolution on the hypergraphs. At the same time, a multi-channel setting is employed in the convolutional network, with each channel encoding its corresponding hypergraph to capture different high-order user relation patterns. In addition, we feed the item embeddings and the obtained high-order user embeddings into a hyperbolic graph convolutional network to extract user and item representations, enabling the model to better capture the hierarchical structure of their complex relationships. Experimental results on three public datasets, namely FilmTrust, LastFM, and Yelp, demonstrate that the model achieves more comprehensive user and item representations, more accurate fitting and processing of graph data, and effectively addresses the issues of insufficient user relationship extraction and data embedding distortion in social recommendation models.

Albumin: a comprehensive review and practical guideline for clinical use.

Nowadays, it is largely accepted that albumin should not be used in hypoalbuminemia or for nutritional purpose. The most discussed indication of albumin at present is the resuscitation in shock states, especially distributive shocks such as septic shock. The main evidence-based indication is also liver disease. In this review, we provided updated evidence-based instruction for definite and potential indications of albumin administration in clinical practice, with appropriate dosing and duration. Data collection was carried out until November 2023 by search of electronic databases including PubMed, Google Scholar, Scopus, and Web of Science. GRADE system has been used to determine the quality of evidence and strength of recommendations for each albumin indication. A total of 165 relevant studies were included in this review. Fluid replacement in plasmapheresis and liver diseases, including hepatorenal syndrome, spontaneous bacterial peritonitis, and large-volume paracentesis, have a moderate to high quality of evidence and a strong recommendation for administering albumin. Moreover, albumin is used as a second-line and adjunctive to crystalloids for fluid resuscitation in hypovolemic shock, sepsis and septic shock, severe burns, toxic epidermal necrolysis, intradialytic hypotension, ovarian hyperstimulation syndrome, major surgery, non-traumatic brain injury, extracorporeal membrane oxygenation, acute respiratory distress syndrome, and severe and refractory edema with hypoalbuminemia has a low to moderate quality of evidence and weak recommendation to use. Also, in modest volume paracentesis, severe hyponatremia in cirrhosis has a low to moderate quality of evidence and a weak recommendation. Albumin administration is most indicated in management of cirrhosis complications. Fluid resuscitation or treatment of severe and refractory edema, especially in patients with hypoalbuminemia and not responding to other treatments, is another rational use for albumin. Implementation of evidence-based guidelines in hospitals can be an effective measure to reduce inappropriate uses of albumin.

Quality appraisal of clinical guidelines for Helicobacter pylori infection and systematic analysis of the level of evidence for recommendations.

To systematically assess the quality of clinical practice guidelines (CPGs) for Helicobacter pylori (HP) infection and identify gaps that limit their development. CPGs for HP infection were systematically collected from PubMed, Embase, the Cochrane Library, the Cumulative Index to Nursing and Allied Health Literature, and six online guideline repositories. Three researchers independently used the AGREE Ⅱ tool to evaluate the methodological quality of the eligible CPGs. In addition, the reporting and recommendation qualities were appraised by using the RIGHT and AGREE-REX tools, respectively. The distribution of the level of evidence and strength of recommendation among evidence-based CPGs was determined. A total of 7,019 records were identified, and 24 CPGs met the eligibility criteria. Of the eligible CPGs, 19 were evidence-based and 5 were consensus-based. The mean overall rating score of AGREE II was 50.7% (SD = 17.2%). Among six domains, the highest mean score was for scope and purpose (74.4%, SD = 17.7%) and the lowest mean score was for applicability (24.3%, SD = 8.9). Only three of 24 CPGs were high-quality. The mean overall score of recommendation quality was 35.5% (SD = 12.2%), and the mean scores in each domain of AGREE-REX and RIGHT were all ≤ 60%, with values and preferences scoring the lowest (16.6%, SD = 11.9%). A total of 505 recommendations were identified. Strong recommendations accounted for 64.1%, and only 34.3% of strong recommendations were based on high-quality evidence. The overall quality of CPGs for HP infection is poor, and CPG developers tend to neglect some domains, resulting in a wide variability in the quality of the CPGs. Additionally, CPGs for HP infection lack sufficient high-quality evidence, and the grading of recommendation strength should be based on the quality of evidence. The CPGs for HP infection have much room for improvement and further researches are required to minimize the evidence gap.

A Collaborative Filtering-Based Recommender Systems approach for Multifarious applications

Recommender systems are crucial in today's IT landscape, enhancing user experiences in various industries. Collaborative filtering (CF) is a key approach, using historical interaction data to predict user preferences. This paper presents CF advancements, with a focus on latent factor models that represent users and items in a compact feature space. It also addresses sparsity issues with techniques like neighborhood-based approaches and content augmentation. Contextual CF, which incorporates temporal and contextual dynamics, is explored through methods like matrix factorization with side information and session-based recommendation. Evaluation metrics such as MAE and RMSE, along with novel ranking-based metrics, provide a comprehensive assessment of recommendation quality. In this paper we outline cutting-edge CF techniques, showcasing their mechanisms and applications and were able to achieve accurate recommendations of almost 90% using MAE and RMSE metrics. By integrating latent factor modeling, sparsity mitigation, contextual enrichment, and advanced evaluation, it paves the way for the next generation of personalized recommendation systems, tailored to meet evolving demands in modern information environments.

The use of guidelines development standards in recommendations and guidelines for retinopathy of prematurity

Objective: Our publication aimed to identify and present standards and tools implemented in the development of guidelines for retinopathy of prematurity (ROP). Methods: The results were based on complementary data extraction from records identified in the systematic review of the guidelines for retinopathy of prematurity performed in 2022. We examined the type of evidence used to develop the guidelines, the standards followed by authors in developing guidelines, and the system for classifying scientific evidence quality. The identified standards for guideline development were checked for validity and the methodology for assessing scientific evidence and presenting recommendations was described. Results: Our analysis of twenty-three retinopathy of prematurity guidelines uncovered two standards: World Health Organization Handbook and Setting Standards for the Development of Clinical Guidelines in Paediatrics and Child Health by Royal College of Paediatrics and Child Health (RCPCH). Only two guidelines from India and Latin America followed the WHO handbook and applied the GRADE system for assessing the quality of evidence and recommendations, while the United Kingdom guidelines adhered to the RCPCH standard and SIGN Grading Hierarchy. Conclusion: Our analysis revealed the need for improvement in the process of creating guidelines for retinopathy of prematurity in most countries. We believe that by raising awareness about the tools available to support authors we are able to spread their use as a standard practice.

A systematic review of nutritional guidelines for preterm infants

The estimated global preterm birth rate in 20201 was more than 10% of livebirths or 13.4 million infants. Nutrition in the neonatal period is a key factor to optimise growth, neurodevelopment, and later metabolic disease risk2. There is no consensus on optimal nutrition for preterm infants, leading to substantial practice variation3. We aimed to assess the quality of nutritional guidelines for preterm infants, the consistency of recommendations, and the gaps in these recommendations. This review is reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) 2020 checklist. The study protocol was registered with PROSPERO (CRD42022327248). We searched six databases and 44 websites for nutritional guidelines for preterm infants before first hospital discharge, which were endorsed, prepared, or authorised by a regional, national, or international body, written in English, and published between 2012-2023. Two reviewers independently screened articles and extracted relevant data including nutritional recommendations (ranges or descriptions), the quality of recommendations (certainty of evidence and strength of recommendation), and gaps in recommendations, defined as those identified by the guidelines or when recommendations were based on very low certainty evidence. Disagreements were resolved by discussion or a third reviewer. Four reviewers appraised the included guidelines using AGREE II. We identified 7051 records, 27 guidelines were included in the review, 26% of which were of high quality. Most guidelines lacked stakeholder involvement and rigour of development. Twelve guidelines had recommendations for macronutrient intake, 18 for micronutrient intake, 12 for feeding, eight for fortification, and 14 for monitoring of nutritional adequacy. Only two guidelines provided recommendations for all five of these aspects. We found considerable variation in recommendations, many of which lacked details of certainty of evidence and strength of recommendation. Recommendations for feeding types and breastmilk fortification were consistent among high quality guidelines, but recommendations varied for intakes of almost all nutrients and monitoring of nutritional adequacy. Different guidelines gave different certainty of evidence for the same recommendations. Most gaps in recommendations were due to a very low certainty of evidence. Future development of nutritional guidelines for preterm infants should follow the standard guideline development method and ensure rigorous process including stakeholders’ involvement to improve the reporting of strength of recommendation, certainty of evidence, and gaps in recommendation. Evidence is needed to support recommendations about macro and micronutrient intakes, breastmilk fortification, and markers on adequacy of intake of different nutrients.

Grading recommendations for enhanced patient safety in sentinel event analysis: the recommendation improvement matrix

ObjectivesThe goal of sentinel event (SE) analysis is to prevent recurrence. However, the rate of SEs has remained constant over the past years. Research suggests this is in part due...

Tackling the cold start issue in movie recommendations with a refined epsilon-greedy approach

With the rapid growth of the Internet and the consequent surge in data, the current era is characterized by information overload. As the domain of data processing and storage expands, recommendation systems have become pivotal tools in navigating this deluge, assisting users in filtering through vast information landscapes. A notable segment of this is movie recommendation systems. As living standards rise, so does the demand for cinematic experiences. Enhancing and refining the methodologies of these recommendation systems is, therefore, of significant value. However, a consistent challenge is the cold start problem encountered when new users join. Without prior viewing records or preferences, these users pose a dilemma for the system: how to offer relevant recommendations without historical data? Addressing this challenge, this paper proposes a unique method grounded in the N-armed bandit model, introducing an enhanced Epsilon-greedy algorithm specifically designed for movie recommendations for such users. By adjusting dynamically based on real-time user feedback, the algorithm aims to continuously hone its recommendation quality, ensuring a consistently better user experience.

An Effective Doctor Recommendation Algorithm for Online Healthcare Platforms

The emergence of online healthcare platforms provides patients with convenience, but choosing the right doctor among the thousands of doctors available on these platforms has become a challenge for patients. The majority of these platforms recommend the same doctors to all patients based on a global ranking, disregarding individual patient preferences. The use of recommender systems helps to resolve this issue by assisting patients in locating doctors who meet their preferences and requirements. Particularly, Collaborative Filtering (CF) algorithms have been extensively utilized to generate personalized recommendations for a variety of applications. Despite their success, they still need to be further optimized to address both the sparsity and cold-start problems due to insufficient data. In this paper, we propose an effective doctor recommendation approach to assist patients in searching for satisfactory doctors who precisely match their preferences regardless of time and location. The proposed approach employs Multi-Criteria CF and content filtering to enhance the quality of recommendations by mitigating the impact of data sparsity and cold start challenges. Offline tests conducted on a real-world dataset show that the proposed approach is superior to state-of-the-art approaches in addressing the aforementioned issues and boosting prediction accuracy and coverage.

Do patients with different craniofacial patterns have differences in upper airway volume? A systematic review with network meta-analysis.

Craniofacial skeletal discrepancies have been associated with upper airway dimensions. To identify differences in upper airway volume across different sagittal and vertical skeletal patterns. Unrestricted literature searches in eight databases/registers for human studies until May 2023. Cross-sectional studies measuring upper airway volumes using three-dimensional imaging in healthy patients of different sagittal (Class I, Class II, and Class III) or vertical (normodivergent, hypodivergent, and hyperdivergent) craniofacial morphology. Duplicate independent study selection, data extraction, and risk of bias assessment. Random-effects frequentist network meta-analysis was performed followed by subgroup-analyses and assessment of the quality of clinical recommendations (confidence in effect estimates) with the CINeMA (Confidence in Network Meta-Analysis) approach. Seventy publications pertaining to 66 unique studies were included with 56 studies (5734 patients) contributing to meta-analyses. Statistically significant differences were found for total pharyngeal airway volume, with Class II having decreased airway volume (-2256.06mm3; 95% Confidence Interval [CI] -3201.61 to -1310.51mm3) and Class III increased airway volume (1098.93mm3; 95% CI 25.41 to 2172.45mm3) compared to Class I. Significant airway volume reductions for Class II were localized mostly at the oropharynx, followed by the palatopharynx, and the glossopharynx. Significant airway volume increases for Class III were localized mostly at the oropharynx, followed by the intraoral cavity, and hypopharynx. Statistically significant differences according to vertical skeletal configuration were seen only for the oropharynx, where hyperdivergent patients had reduced volumes compared to normodivergent patients (-1716.77mm3; 95% CI -3296.42 to -137.12mm3). Airway differences for Class II and Class III configurations (compared to Class I) were more pronounced in adults than in children and the confidence for all estimates was very low according to CINeMA. Considerable differences in upper airway volume were found between sagittal and vertical skeletal configurations. However, results should be interpreted with caution due to the high risk of bias, owing to the retrospective study design, inconsistencies in anatomic compartment boundaries used, samples of mixed children-adult patients, and incomplete reporting. PROSPERO (CRD42022366928).

Leveraging Opposite Gender Interaction Ratio as a Path towards Fairness in Online Dating Recommendations Based on User Sexual Orientation

Online dating platforms have gained widespread popularity as a means for individuals to seek potential romantic relationships. While recommender systems have been designed to improve the user experience in dating platforms by providing personalized recommendations, increasing concerns about fairness have encouraged the development of fairness-aware recommender systems from various perspectives (e.g., gender and race). However, sexual orientation, which plays a significant role in finding a satisfying relationship, is under-investigated. To fill this crucial gap, we propose a novel metric, Opposite Gender Interaction Ratio (OGIR), as a way to investigate potential unfairness for users with varying preferences towards the opposite gender. We empirically analyze a real online dating dataset and observe existing recommender algorithms could suffer from group unfairness according to OGIR. We further investigate the potential causes for such gaps in recommendation quality, which lead to the challenges of group quantity imbalance and group calibration imbalance. Ultimately, we propose a fair recommender system based on re-weighting and re-ranking strategies to respectively mitigate these associated imbalance challenges. Experimental results demonstrate both strategies improve fairness while their combination achieves the best performance towards maintaining model utility while improving fairness.

FacetCRS: Multi-Faceted Preference Learning for Pricking Filter Bubbles in Conversational Recommender System

The filter bubble is a notorious issue in Recommender Systems (RSs), which describes the phenomenon whereby users are exposed to a limited and narrow range of information or content that reinforces their existing dominant preferences and beliefs. This results in a lack of exposure to diverse and varied content. Many existing works have predominantly examined filter bubbles in static or relatively-static recommendation settings. However, filter bubbles will be continuously intensified over time due to the feedback loop between the user and the system in the real-world online recommendation. To address these issues, we propose a novel paradigm, Multi-Facet Preference Learning for Pricking Filter Bubbles in Conversational Recommender System (FacetCRS), which aims to burst filter bubbles in the conversational recommender system (CRS) through timely user-item interactions via natural language conversations. By considering diverse user preferences and intentions, FacetCRS automatically model user preference into multi-facets, including entity-, word-, context-, and review-facet, to capture diverse and dynamic user preferences to prick filter bubbles in the CRS. It is an end-to-end CRS framework to adaptively learn representations of various levels of preference facet and diverse types of external knowledge. Extensive experiments on two publicly available benchmark datasets demonstrate that our proposed method achieves state-of-the-art performance in mitigating filter bubbles and enhancing recommendation quality in CRS.

Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework

Recommender systems have a significant impact on various real-world applications, shaping people's daily lives and enhancing productivity. Traditional recommender models aim to collect extensive user information to accurately estimate user preferences. However, in practical scenarios, users may not want all their behaviors to be included in the model training process. This paper introduces a novel recommendation paradigm that allows users to indicate their ``willingness'' regarding which data should contribute to model training. The models are then optimized to maximize utility, which considers the trade-off between recommendation performance and respecting user preferences. The recommendation problem is formulated as a multiplayer game, with each user acting as a player and using a selection vector to indicate their willingness to include specific interacted items in training. To efficiently solve this game, an influence function-based model is proposed to approximate recommendation performances for different actions without re-optimizing the model. Furthermore, an enhanced model leveraging multiple anchor actions for the influence function is introduced to improve performance approximation accuracy. The convergence rate of the algorithm is theoretically analyzed, and the advantages of incorporating multiple anchor actions are demonstrated. Extensive experiments on both simulated and real-world datasets validate the effectiveness of the proposed models in balancing recommendation quality and user willingness. To promote this research direction, we have released our project at https://paitesanshi.github.io/IFRQE/.

Improving the methord of collaborative filtering by integrating semantic and temporal factors and the methord of cluster analysis.

The article examines the algorithm for generating recommendations based on collaborative filtering, taking into account the influence of semantic and time factors and its improvement using cluster analysis methods in order to reduce the load on the recommendation system and improve the quality of recommendations by filtering out meaningless content and preserving the context during the generation of recommendations. The impact of semantic and time factors on the quality of the recommendation system (error in estimation approximation) and the application of the cluster analysis method on the speed of the system with a large set of data are analyzed. A technique for accelerating the processing of received data about users is proposed, which consists in an attempt to take into account the fact that users' interests change over time and the possibility of breaking down the content of statistical data by a set of specific features. A data preprocessing procedure (data aggregation) was formulated for the method of collaborative filtering based on comparisons of objects using the clustering method, which made it possible to reduce the complexity of calculations and, accordingly, the time for the formation of recommendations. An algorithm for calculating the object's assessment is presented, taking into account temporal and semantic factors. The software was developed, the adequacy of the proposed method was verified using data sets from different domain areas. As a result of the verification, it was found that the modified algorithm has better performance indicators compared to the naive method

A hybrid similarity model for mitigating the cold-start problem of collaborative filtering in sparse data

Similarity is a vital component for neighborhood-based collaborative filtering (CF). To improve the quality of recommendation, many similarity methods have been proposed and analyzed in recent decades. However, nearly all traditional similarity methods and many advanced similarity methods only utilize corated items among users to compute their similarity, which provides limited information in cold-start/sparse scenarios and yields misleading results. In addition, although a few advanced hybrid similarity models consider items beyond corated items, which can partly mitigate the above limitation, they still have drawbacks, such as disregarding penalizing noncorated items that have many disadvantages. In this paper, we explore a new robust hybrid similarity model, namely Wasserstein distance-based CF (WCF) model, for mitigating the cold-start problem of CF in sparse data. Specifically, we measure item similarity via the Wasserstein distance, which can help circumvent the drawbacks in the Bhattacharyya coefficient and KL divergence that are used in the literature, and is thus more robust in a cold-start/sparse scenario. Besides, we further design a new multiplicative user similarity formula which identifies all noncorated items as a whole to prioritize the importance of corated items and impair the negative effects of noncorated items, which will also play an important role in a cold-start/sparse scenario. In addition, we also propose two novel heuristic similarity factors to impair the negative effects of popular users and items as supplements. We conduct extensive experiments on five real-world benchmark recommendation datasets to test WCF. The experimental results show the superiority of WCF over other existing similarity methods in cold-start/sparse scenarios.

An Improved Fusion-Based Semantic Similarity Measure for Effective Collaborative Filtering Recommendations

Semantic-enhanced recommendation systems are promising approaches to overcome the sparsity and cold-start problems, which are hard to handle using the conventional collaborative filtering (CF) approaches. Further research is needed to effectively integrate ontologies into collaborative filtering recommender systems. This paper proposes an ontology-based semantic similarity measure to evaluate similarities between items and eventually generate accurate recommendations. The proposed semantic similarity measure termed fusion-based semantic similarity takes into account the semantics of ontological instances (i.e. items) inferred from a specific domain ontology, which is determined by analyzing the hierarchical relationships among the instances, as well as the features of the instances and their relationships to other instances. The new measure comprehensively captures the semantic knowledge associated with instances by exploiting all possible shared semantics between instances in a given domain ontology. Furthermore, this paper proposes a new semantic-enhanced hybrid recommendation approach as a result of combining the new semantic similarity measure with the standard item-based CF to enhance the quality of generated recommendations. In order to assess the effectiveness of our semantic-enhanced hybrid collaborative filtering method, a series of experiments were conducted to compare the performance of the proposed approach against well-established benchmark techniques. The reported experimental results consistently emphasize its superiority, demonstrating enhanced predictive abilities and a notable improvement in the quality of recommendations. More specifically, the proposed approach achieved notable 6% reduction in Mean Absolute Error (MAE) in certain cases, outperforming other benchmark techniques. Additionally, this study highlights the potential of using semantic-based similarity to enhance the performance of recommendation systems. Such enhancements address challenges within collaborative filtering, potentially leading to advancements in recommendation system design and optimization.

Hybrid knowledge-infused collaborative filtering for enhanced movie clustering and recommendation

This article proposes an enhanced knowledge-based collaborative filtering model for movie recommendation services to address the limitations of collaborative filtering in capturing the diverse preferences and specific characteristics of movies. The proposed model integrates external knowledge sources, such as movie plots and reviews, to enrich the recommendation process. By leveraging this additional information, the model can better understand movies’ unique features and attributes, improving recommendation accuracy and relevance. The knowledge-based features are extracted and incorporated into the collaborative filtering framework, enhancing the model’s ability to match user preferences with movie characteristics. Experiments are conducted using the MovieLens dataset to evaluate the proposed model. The MAE and RMSE metrics are employed to assess the quality of recommendations. Comparative analyses are conducted against various baseline approaches, including popularity-based, CF-based, content-based, and hybrid recommendation models. The experimental results demonstrate the effectiveness of the proposed knowledge-based collaborative filtering model. The proposed model consistently outperforms the baselines, providing more accurate and personalized recommendations.

Multi-Channel Hypergraph Collaborative Filtering with Attribute Inference

In the field of collaborative filtering, attribute information is often integrated to improve recommendations. However, challenges remain unaddressed. Firstly, existing data modeling methods often fall short of appropriately handling attribute information. Secondly, attribute data are often sparse and can potentially impact recommendation performance due to the challenge of incomplete correspondence between the attribute information and the recommendations. To tackle these challenges, we propose a hypergraph collaborative filtering with attribute inference (HCFA) framework, which segregates attribute and user behavior information into distinct channels and leverages hypergraphs to capture high-order correlations among vertices, offering a more natural approach to modeling. Furthermore, we introduce behavior-based attribute confidence (BAC) for assessing the reliability of inferred attributes concerning the corresponding behaviors and update the most credible portions to enhance recommendation quality. Extensive experiments conducted on three public benchmarks demonstrate the superiority of our model. It consistently outperforms other state-of-the-art approaches, with ablation experiments further confirming the effectiveness of our proposed method.