Incorporating user experience into critiquing-based recommender systems: a collaborative approach based on compound critiquing

With the increasing problem of information overload, recommenders are essential on many websites to help users find interesting content quickly. User feedback is the most valuable form of model feedback, and the form and how user feedback is used affects the accuracy, coverage and experience of recommendations. Thus, this paper reviews feedback types in recommender systems: explicit, implicit, and deliberate implicit feedback, and discusses their pros and cons in terms of accuracy, ease-of-use, and user effort. Additionally, it discusses challenges recommendation models face when using feedback, like data sparsity, bias reinforcement, and fairness limitations. Meanwhile, it proposes strategies to address these challenges by integrating multi-source feedback, fairness-aware modeling, and interpretability enhancement to boost user confidence. The results show gains in accuracy, coverage, and fairness. In addition, it provides exploratory analyses on dynamic feedback and cross-domain recommendations, offering insights for modeling and applying multi-type feedback in future recommender systems.

Learning user preferences by modeling historical purchase behaviors has significantly succeeded in existing recommender systems. Most use trained models to make predictions for users, and they assume that the training data samples and test data sample sets come from the same distribution. However, in practical applications, the distribution of users’ true preferences may be more complicated, and data drift can easily make the trained model invalid on the test dataset. In this case, to accurately model user preferences based on their historical behavior, two difficulties need to be addressed. First, it is difficult to model various purchase behavior shift situations due to their complexity. Second, inferring users’ true preferences from the complicated shifting cases is challenging. To solve the above problems, we build a robust recommender system to predict possible user purchase shifts and make recommendations for users. First, we propose a simulating strategy to cover possible scenarios when user purchase behavior shifts. Second, we build a novel voting framework to ensure the robustness of predicting results based on learned preferences. Extensive experiments were conducted, and the results demonstrate the outstanding performance of the proposed method on MovieLens-1M and LastFM datasets, providing at most 37.31% and 30.95% relative performance gains, respectively.

Preference learning plays a crucial role in AI applications, particularly in recommender systems and personalized services. Traditional pairwise comparisons, while foundational, present scalability challenges in large-scale systems. This study explores alternative elicitation methods such as ranking, numerical ratings, and natural language feedback, alongside a novel hybrid framework that dynamically integrates these approaches. The proposed methods demonstrate improved efficiency, reduced cognitive load, and enhanced accuracy. Results from simulated user studies reveal that hybrid approaches outperform traditional methods, achieving a 40% reduction in user effort while maintaining high predictive accuracy. These findings open pathways for deploying user-centric, scalable preference learning systems in dynamic environments.

Due to the dynamic and heterogeneous nature of the Web, it is increasingly difficult for users to choose amongst large amounts of data. For this reason, the modelling of users and access to personalised information becomes essential. Recommendation systems stand out as systems that aim to provide the most appropriate and efficient services by offering personalised recommendations to users. Traditional recommendation systems provide suggestions to their users with static approaches and do not include user preferences that change over time in suggestion strategies. In this study, a comprehensive review and comparison of recommendation systems that can adaptively develop suggestion approaches according to changing user preferences and learn user preferences are presented.KeywordsCollaborative filteringClassification techniqueRecommendation systemSocial mediaUser preferences

With the popular and increasing power of the Internet these days, the effort of distributing and inventory costs of stocking various online retailing items are nearly negligible. In addition to selling popular, called short-head, items in large quantities, online retailers, such as Amazon, offer a large number of unique items, called long tail, with relatively small quantities sold. Retailers realize that it has high value to sell items from the long-tail category, since for users these long-tail items could meet the interest of them and surprise them simultaneously. Retailers also recognize that long-tail items can be an untapped source of revenue for a business; however, it is difficult to connect customers with long-tail items they are interested in, since they are unaware of them. Recommender systems help bridge the gap between users and long-tail items by learning user preferences and recommending appropriate items to them. In this paper, we propose a new tripartite graph recommender system, which is designed to suggest long-tail items. Compared with other graph-based recommender systems, our proposed recommendation system solves the tripartite variant problem suffered by existing approaches for having a low diversity score. A rework of the tripartite graph system is introduced, called the extended tripartite graph system, which enhances the performance of existing long-tail recommendation approaches measured by using two widely-used performance metrics: recall and diversity. Experimental results on the extended tripartite graph algorithm verify its merits and novelty.

Recommendation systems provide users with personalized recommendations, stand out as systems aiming to provide appropriate and efficient services. Traditional recommendation systems provide suggestions to their users with static approaches and do not include user preferences that change over time in suggestion strategies. In this study, a comprehensive review and comparison of recommendation systems that can adaptively develop suggestion approaches according to changing user preferences and learn user preferences is presented. In this research other approaches and solutions with frameworks of Deep Reinforcement Learning will be compared. In all articles that is reviewed Markov Decision Process (MDP) were used as a solution for dynamic recommendations and long-term rewards for the users. Proposed frameworks like DEERS, DRR etc. will be reviewed.

Large language model (LLM) agents have emerged as a promising solution for enhancing recommendation systems via user simulation. However, existing studies predominantly resort to prompt-based simulation using frozen LLMs, which frequently results in suboptimal item modeling and user preference learning, thereby ultimately constraining recommendation performance. To address these challenges, we introduce VRAgent-R1, a novel agent-based paradigm that incorporates human-like intelligence in user simulation. Specifically, VRAgent-R1 comprises two distinct agents: the Item Perception (IP) Agent and the User Simulation (US) Agent, designed for interactive user-item modeling. Firstly, the IP Agent emulates human-like progressive thinking based on MLLMs, effectively capturing hidden recommendation semantics in videos. With a more comprehensive multimodal content understanding provided by the IP Agent, the video recommendation system is equipped to provide higher-quality candidate items. Subsequently, the US Agent refines the recommended video sets based on in-depth chain-of-thought (CoT) reasoning and achieves better alignment with real user preferences through reinforcement learning. Experimental results on a large-scale video recommendation benchmark MicroLens-100k have demonstrated the effectiveness of our proposed VRAgent-R1 method, e.g., the IP Agent achieves a 6.0% improvement in NDCG@10, while the US Agent shows approximately 45.0% higher accuracy in user decision simulation compared to state-of-the-art baselines.

If properly analyzed, the multi-aspect rating data could be a source of rich intelligence for providing personalized restaurant recommendations. Indeed, while recommender systems have been studied for various applications and many recommendation techniques have been developed for general or specific recommendation tasks, there are few studies for restaurant recommendation by addressing the unique challenges of the multiaspect restaurant reviews. As we know, traditional collaborative filtering methods are typically developed for single aspect ratings. However, multi-aspect ratings are often collected from the restaurant customers. These ratings can reflect multiple aspects of the service quality of the restaurant. Also, geographic factors play an important role in restaurant recommendation. To this end, in this paper, we develop a generative probabilistic model to exploit the multi-aspect ratings of restaurants for restaurant recommendation. Also, the geographic proximity is integrated into the probabilistic model to capture the geographic influence. Moreover, the profile information, which contains customer/restaurantindependent features and the shared features, is also integrated into the model. Finally, we conduct a comprehensive experimental study on a real-world data set. The experimental results clearly demonstrate the benefit of exploiting multi-aspect ratings and the improvement of the developed generative probabilistic model.

Nowadays, graph collaborative filtering is the most practical and classical method for recommender systems, which learns user preferences through historical interactions between users and items. The recent combination of contrastive learning and recommender systems has led to a continuous improvement in the performance of graph collaborative filtering. A variety of data augmentation methods have also emerged. However, some graph augmentation methods are proven to lose graph information, such as the way of dropout on the graph structure. Inspired by different adversarial samples and different data augmentation methods, we adopt the method of adding interference patterns to the embedding space, as opposed to other models, we take a partial sampling of the embedding and then add the partially sampled data into the embedding, thus forming different sub-views. The extensive experimental results show that our experiments can improve the performance of the model more compared to other contrastive learning methods with our data enhancement method.

In the area of knowledge-based information systems, TV recommendation systemsRecommendation systems have attracted researchers’ attention with the development of Smart TVSmart TV and expansion of TV contentTV content . Traditional TV recommendation systems, which are based on individual’s viewing activity and only recommending TV programs, have been unable to meet the requirements of Smart TV. Hence, this chapter proposes a hybrid recommendation system, which not only uses the information of single user’s activity, but also takes into account other users’ viewing habits and related information from the Internet, for different content such as TV programs, movies, and music. The proposed recommendation system integrates Content Analysis Component, User Analysis Component and Preference Learning Component. Moreover, this chapter also discusses several important design issues, such as diversity, novelty, explanation and group recommendations, which should be considered in designing/building a TV recommendation system. The proposed framework could be used to help designers and developers to design a TV recommendation system engine for Smart TV.

The use of social robots in the domestic environment has increased during the past few decades. These robots are intended to maintain long-term interactions with humans while involving in a variety of tasks including daily activities, entertainment, and assisting elderly or disabled people. The ability to learn user's preferences and adapting interaction accordingly is a must for such robots. As the domestic environment consists of non-experts, social robots must possess more natural and human-friendly interaction capabilities. This paper presents an Autobiographical Memory(AM) based intelligent system which can learn user preferences through natural interactions and provide user adaptive services for each user in the multi-user domestic environment. The system is capable of learning user's preferences from hislher own statements and from another person's statements. Furthermore, the system is capable of adapting to user's hidden preference and changes of preferences easily. The robot's memory has been structured in such a way that it can easily remember the user groups and the relationship between users. This facilitates the robot for learning preferences which are common to a group of users. The system has been tested and validated using snack and beverage suggestion scenario.

Recommender systems have become valuable resources for users seeking intelligent ways to search through the enormous volume of information available to them. One crucial unsolved problem for recommender systems is how best to learn about a new user. In this paper we study six techniques that collaborative filtering recommender systems can use to learn about new users. These techniques select a sequence of items for the collaborative filtering system to present to each new user for rating. The techniques include the use of information theory to select the items that will give the most value to the recommender system, aggregate statistics to select the items the user is most likely to have an opinion about, balanced techniques that seek to maximize the expected number of bits learned per presented item, and personalized techniques that predict which items a user will have an opinion about. We study the techniques thru offline experiments with a large pre-existing user data set, and thru a live experiment with over 300 users. We show that the choice of learning technique significantly affects the user experience, in both the user effort and the accuracy of the resulting predictions.

This chapter examines the phenomenon of internet users attempting to report and prevent online child sexual exploitation (CSE) and child sexual abuse material (CSAM) in the absence of adequate intervention by internet service providers, social media platforms, and government. The chapter discusses the history of online CSE, focusing on regulatory stances over time in which online risks to children have been cast as natural and inevitable by the hegemony of a “cyberlibertarian” ideology. We illustrate the success of this ideology, as well as its profound contradictions and ethical failures, by presenting key examples in which internet users have taken decisive action to prevent online CSE and promote the removal of CSAM. Rejecting simplistic characterizations of “vigilante justice,” we argue instead that the fact that often young internet users report feeling forced to act against online CSE and CSAM undercuts libertarian claims that internet regulation is impossible, unworkable, and unwanted. Recent shifts toward a more progressive ethos of online harm minimization are promising; however, this ethos risks offering a new legitimizing ideology for online business models that will continue to put children at risk of abuse and exploitation. In conclusion, we suggest ways forward toward an internet built in the interests of children, rather than profit.

The matrix factorization (MF) technique has been widely adopted for solving the rating prediction problem in recommender systems. The MF technique utilizes the latent factor model to obtain static user preferences (user latent vectors) and item characteristics (item latent vectors) based on historical rating data. However, in the real world, user preferences are not static but full of dynamics. Though there are several previous works that addressed this time-varying issue of user preferences, it seems (to the best of our knowledge) that none of them are specifically designed for tracking concept drift in individual user preferences . Motivated by this, we develop a temporal MF approach for tracking concept drift in each individual user latent vector. There are two key innovative steps in our approach: 1) we develop a modified stochastic gradient descent method to learn an individual user latent vector at each time step and 2) by Lasso regression, we learn a linear model for the transition of the individual user latent vectors. We test our method on a synthetic data set and several real data sets. In comparison with the original MF, our experimental results show that our temporal method is able to achieve lower root mean square errors (RMSEs) for both the synthetic and real data sets. One interesting finding is that the performance gain in RMSE is mostly from those users who indeed have concept drift in their user latent vectors at the time of prediction. In particular, for the synthetic data set and the Ciao data set, there are quite a few users with that property and the performance gains for these two data sets are roughly 20% and 5%, respectively.

Recommender systems have become increasingly accurate in suggesting content to users, resulting in users primarily consuming content through recommendations. This can cause the user's interest to narrow toward the recommended content, something we refer to as preference amplification. While this can contribute to increased engagement, it can also lead to negative experiences such as lack of diversity and echo chambers. We propose a theoretical framework for studying such amplification in a matrix factorization based recommender system. We model the dynamics of the system, where users interact with the recommender systems and gradually drift'' toward the recommended content, with the recommender system adapting, based on user feedback, to the updated preferences. We study the conditions under which preference amplification manifests, and validate our results with simulations. Finally, we evaluate mitigation strategies that prevent the adverse effects of preference amplification and present experimental results using a real-world large-scale video recommender system showing that by reducing exposure to potentially objectionable content we can increase user engagement by up to 2%.