Research On Multi-Domain Recommendation Systems Based on The Thompson Sampling Algorithm

With the advent of the digital economy era, business systems such as web advertising and recommendation system have put forward the demand for predicting the click through rate (CTR) of items. However, the current CTR prediction research is not enough to mine user behavior, resulting in the lack of accuracy of user interest representation. In this paper, we propose a CTR prediction model, called MLIM, which can deep mine the evolution law of user interest. Specifically, we first use BiGRU to obtain the low-level user interest representation in the interest extraction layer, and then continue to use attention mechanism, BiGRU and sliding time window multi-components collaborative modeling in the interest evolution layer to obtain multi-level user interest representation with richer information, which can improve the accuracy of CTR prediction to a certain extent. Comprehensive experiments on two real datasets show that the proposed model achieves better performance than the mainstream baselines integrating user behavior analysis.

Recommender systems play an important role in most modern e-commerce applications. They have allowed users to become aware of the myriad choices available to them. The ease of information and the abundance of options have helped users make educated decisions. A recommender system studies a user's preferences and continues learning the user's changing interests, so as to suggest items that incline with the user's interests. In cases where a user is new to the application, or the user prefers not to discourse preferences, the recommender system is unable to gather the user's preference on any item. This is called the cold start problem; wherein the system can make valid recommendations only once the user starts informing the system about his/her choices. In this paper, we discuss the challenges faced by the cold start problem and how this problem may be alleviated using social media. We suggest an approach where we collect public information from users' social media accounts and analyze this information to understand their preferences. In particular, we gather the new user's information using their Twitter profile; i.e., the user's interest and preferences are extracted from his/her Twitter profile by analyzing his/her tweets. These interests will help the system understand what kind of movies the user will be most interested in. We compare these preferences with the metadata about the individual items. Using this approach, we develop a movie recommendation system wherein we produce top-N movie recommendations for a user. We used the MovieTweetings dataset to model the application. Two sets of results have been produced. In the first, smaller set of 770 users, 72.67% of users have received 100% accurate movie recommendations while nearly 80% of users got more than 75% accuracy. For the second, larger set of more than 3,500 users, 53 % of users have received 100% accurate recommendations while 72% of users got more than 75% accuracy. These encouraging results have demonstrated that the approach is effectively in alleviating cold start problems in recommendation systems, and may be applicable to many other e-commerce applications.

Click-through rate (CTR) prediction plays a critical role in recommender systems and other applications. Recently, modeling user behavior sequences attracts much attention and brings great improvements in the CTR field. Many existing works utilize attention mechanism or recurrent neural networks to exploit user interest from the sequence, but fail to recognize the simple truth that a user's real-time interests are inherently diverse and fluid. In this paper, we propose DisenCTR, a novel dynamic graph-based disentangled representation framework for CTR prediction. The key novelty of our method compared with existing approaches is to model evolving diverse interests of users. Specifically, we construct a time-evolving user-item interaction graph induced by historical interactions. And based on the rich dynamics supplied by the graph, we propose a disentangled graph representation module to extract diverse user interests. We further exploit the fluidity of user interests and model the temporal effect of historical behaviors using Mixture of Hawkes Process. Extensive experiments on three real-world datasets demonstrate the superior performance of our method comparing to state-of-the-art approaches.

Recommender systems play a critical role in e-commerce, streaming media, and other fields, where user engagement is vital to platform success. These systems, however, face challenges like the cold start problem, which limits recommendation accuracy for new users and items due to sparse data. Although multi-armed bandit (MAB) algorithms are considered effective for balancing exploration and exploitation, existing studies lack simulations of algorithm performance in scenarios where recommender systems indirectly influence user choices, limiting insights for performance improvements in real-world applications. This study explores the performance of key MAB algorithmsExplore-Then-Commit (ETC), Upper Confidence Bound (UCB), and Thompson Sampling (TS)under a novel framework that models user behavior as an unreliable intermediate using Markov state transitions. This Markov perturbation-based approach provides a foundation for indirectly influencing user decisions. Two algorithmic frameworks are examined: one considering only the recommendation system's influence on user choices, and another also considering user feedback as a dynamic influence on state transition probabilities. Results show that, while considering user feedback increases adaptability, it also causes greater volatility. TS achieves the lowest cumulative regret with strong stability, while Asymptotic-UCB achieves a better balance between exploration and exploitation than standard UCB, with consistent adaptability under varying feedback. In contrast, ETC demonstrates significant volatility, highlighting its limitations in feedback-driven scenarios. These findings provide valuable insights for enhancing MAB algorithms in real-world cold start recommendation applications, especially where systems influence user choices indirectly.

Collaborative filtering (CF), a technique used by recommendation systems, predicts and recommends items (information, products or services) that the user might like. Amazon.com's recommender system is one of the most famous examples of CF. Recommendation systems are popular in both commercial and research sectors, and they are applied in a variety of applications such as movies, music, books, social connections and venues. In particular, movie recommendation systems produce personal recommendations for movies. Existing CF algorithms employed in movie recommendation systems predict the unknown rating of a given user for a movie using only the ratings (i.e., preferences) of other like-minded users who have seen the movie. In such approaches, there exist certain limits in improving the accuracy of recommendation systems. This paper proposes an algorithm for movie recommendation that exploits the genre of the movie to enhance the accuracy of rating predictions. The proposed algorithm 1) numerically measures the correlation between movie genres using movie rating information; 2) classifies movies using the genre correlations and generates a list of recommended movies for the target user with the classified movies; and finally 3) predicts the ratings of the movies in the list using traditional CF algorithms. The experimental results show that the proposed algorithm yields higher accuracy in movie rating predictions than existing movie recommendation algorithms.

In an age where digital connectivity is increasingly shaping entertainment content, personalized movie recommendations play a pivotal role in enhancing user satisfaction and engagement. This research introduces an innovative approach utilizing Enhanced Self-Organizing Maps (SOM) to streamline movie selection processes. Self-Organizing Maps (SOMs), a type of unsupervised neural network architecture, are particularly adept at discerning intricate data patterns, making them valuable assets in recommendation systems. The methodology outlined in this paper commences with gathering user-movie interaction data, including user feedback and movie characteristics, which is standardized to ensure consistency before model training. Leveraging its adaptable learning rate and neighborhood function, the Enhanced SOM effectively identifies subtle data nuances. Personalized movie suggestions are then generated by exploiting the Enhanced SOM's capacity to identify similar users and films. Integration of hybrid filtering techniques enriches recommendation quality, blending collaborative filtering algorithms, which leverage user-item interactions, with content-based filtering, which utilizes movie attributes such as genres and descriptions. This amalgamation results in suggestions that harmoniously combine diverse filtering methodologies. The proposed solution's efficacy is rigorously evaluated by comparing suggestion accuracy and user satisfaction against predefined benchmarks. Extensive real-world dataset testing corroborates the effectiveness of the Enhanced SOM-based movie recommendation approach. Furthermore, the system offers flexibility through options for parameter adjustment, grid size variations, and neighborhood function modifications to further refine recommendation accuracy. Collectively, these elements underscore the efficacy of the proposed method in furnishing tailored movie recommendations. When coupled with hybrid filtering techniques, the implementation of Enhanced SOMs emerges as a reliable model for content platforms seeking to enhance user experiences by delivering precise movie recommendations, coupled with scalability and adaptability.

Recommender Systems are mainly used in various e-commerce applications, especially online stores threatening users’ privacy. The privacy issues can be overcome by using security solutions, which include blockchain technology for privacy applications. The fusion of the Internet of Things and blockchain technology has fully improved modern distributed systems. The combination guarantees the safety and scalability of the recommender system. We aim to create an authorized secure exchange device using blockchain-enabled multiparty computation by adding smart contracts to the core blockchain protocol. The recommendation structure and Blockchain technology make online shopping more convenient and private. We propose a blockchain-related recommender system using the “movielens” data. The case study includes a smart contract model that recommends movies to buyers. Initially, we tested the model on a small “movielens dataset” and extended it to a 3M movielens dataset. We developed a classifier model for movielens and proposed a Dual light graph convolutional network for movielens data classification. Our results, including ablation analysis, show that blockchain strategies and Dual light graph convolutional networks can effectively improve recommender systems’ privacy. Furthermore, the suggested blockchain technique can be stretched by similar procedures.

CTR (Click-Through Rate) prediction, crucial for recommender systems and online advertising, etc., has been confirmed to benefit from modeling long-term user behaviors. Nonetheless, the vast number of behaviors and complexity of noise interference pose challenges to prediction efficiency and effectiveness. Recent solutions have evolved from single-stage models to two-stage models. However, current two-stage models often filter out significant information, resulting in an inability to capture diverse user interests and build the complete latent space of user interests. Inspired by multi-interest and generative modeling, we propose DiffuMIN (Diffusion-driven Multi-Interest Network) to model long-term user behaviors and thoroughly explore the user interest space. Specifically, we propose a target-oriented multi-interest extraction method that begins by orthogonally decomposing the target to obtain interest channels. This is followed by modeling the relationships between interest channels and user behaviors to disentangle and extract multiple user interests. We then adopt a diffusion module guided by contextual interests and interest channels, which anchor users' personalized and target-oriented interest types, enabling the generation of augmented interests that align with the latent spaces of user interests, thereby further exploring restricted interest space. Finally, we leverage contrastive learning to ensure that the generated augmented interests align with users' genuine preferences. Extensive offline experiments are conducted on two public datasets and one industrial dataset, yielding results that demonstrate the superiority of DiffuMIN. Moreover, DiffuMIN increased CTR by 1.52% and CPM by 1.10% in online A/B testing. Our source code is available at https://github.com/laiweijiang/DiffuMIN.

The user, usage, and usability (3U’s) are three principal constituents for cyber security. The effective analysis of the 3U data using artificial intelligence (AI) techniques allows to deduce valuable observations, which allow domain experts to design practical strategies to alleviate cyberattacks and ensure decision support. Many internet applications, such as internet advertising and recommendation systems, rely on click-through rate (CTR) prediction to anticipate the possibility that a user would click on an ad or product, which is key for understanding human online behaviour. However, online systems are prone to click on fraud attacks. We propose a Human-Centric Cyber Security (HCCS) model that additionally includes AI techniques targeted at the key elements of user, usage, and usability. As a case study, we analyse a CTR prediction task, using deep learning methods (factorization machines) to predict online fraud through clickbait. The results of experiments on a real-world benchmark Avazu dataset show that the proposed approach outpaces (AUC is 0.8062) other CTR forecasting approaches, demonstrating the viability of the proposed framework.

The necessity for sophisticated recommender systems in the movie recommendation sphere has become particularly pronounced, generating a more personalized movie recommendation due to people nowadays who like to watch movies online. Efficient recommender systems make use of advanced machine learning (ML) techniques in the pursuit of accurate and meaningful recommendations. This paper endeavours to give a comprehensive overview of technologies known as recommender systems, concentrating on ML methods found at the base. Different strategies have been applied in this work, which include collaborative filtering (CF), content-based filtering (CB), hybrid approaches, Generative AI and so on. The merits and demerits of each technique are listed and explained briefly. In addition, the actual application’s results are also presented in this paper. To evaluate the performance of the techniques, some of the important datasets that are used in evaluating recommender systems are also discussed along with measurement metrics to determine the effectiveness of technique. Example metrics used are Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and so on. This paper synthesizes existing research to evaluate the advantages and limitations of diverse recommendation techniques, which aims to give suggestion on how to improve the design of movie recommendation systems so that the performance of the technique can be improved.

Recently, the demand for online and offline recommendation systems has increased drastically. These systems are widely used in tourism, music, and video or movie recommendations. Currently, online movie streaming applications have gained huge attention. Providing better recommendations to the user is a challenging task for these applications. The content-based filtering (CBF) recommender system is a promising technique for these systems. However, traditional systems suffer from challenges such as cold-start problems, sparsity, and scalability. Consequently, we strengthen content-based recommendation algorithms by enriching the user-related and relevant product models with effective tendencies. The majority of previous work on classifiers has been in recommendation systems. To overcome these issues, we present a deep learning model that uses a deep neural network mechanism and a multiplicative BiLSTM model. This scheme uses embedding, weight updating, and preference learning processes to improve the recommendation system's performance. The performance of the proposed approach is measured in terms of MAE, MAP, Precision, Recall, and F-measure. The comparative performance shows that the proposed approach achieves better performance when compared with state-of-art movie recommendation techniques.

Recommender system is a tool to suggest items to the users from the extensive history of the user's feedback. Though, it is an emerging research area concerning academics and industries, where it suffers from sparsity, scalability, and cold start problems. This paper addresses sparsity, and scalability problems of model-based collaborative recommender system based on ensemble learning approach and enhanced clustering algorithm for movie recommendations. In this paper, an effective movie recommendation system is proposed by Classification and Regression Tree (CART) algorithm, enhanced Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm and truncation method. In this research paper, a new hyper parameters tuning is added in BIRCH algorithm to enhance the cluster formation process, where the proposed algorithm is named as enhanced BIRCH. The proposed model yields quality movie recommendation to the new user using Gradient boost classification with broad coverage. In this paper, the proposed model is tested on Movielens dataset, and the performance is evaluated by means of Mean Absolute Error (MAE), precision, recall and f-measure. The experimental results showed the superiority of proposed model in movie recommendation compared to the existing models. The proposed model obtained 0.52 and 0.57 MAE value on Movielens 100k and 1M datasets. Further, the proposed model obtained 0.83 of precision, 0.86 of recall and 0.86 of f-measure on Movielens 100k dataset, which are effective compared to the existing models in movie recommendation.

LinkedIn is the largest professional social network in the world with more than 238M members. It provides a platform for advertisers to reach out to professionals and target them using rich profile and behavioral data. Thus, online advertising is an important business for LinkedIn. In this talk, I will give an overview of machine learning and optimization components that power LinkedIn self-serve display advertising systems. The talk will not only focus on machine learning and optimization methods, but various practical challenges that arise when running such components in a real production environment. I will describe how we overcome some of these challenges to bridge the gap between theory and practice.The major components that will be described in details include Response prediction: The goal of this component is to estimate click-through rates (CTR) when an ad is shown to a user in a given context. Given the data sparseness due to low CTR for advertising applications in general and the curse of dimensionality, estimating such interactions is known to be a challenging. Furthermore, the goal of the system is to maximize expected revenue, hence this is an explore/exploit problem and not a supervised learning problem. Our approach takes recourse to supervised learning to reduce dimensionality and couples it with classical explore/exploit schemes to balance the explore/exploit tradeoff. In particular, we use a large scale logistic regression to estimate user and ad interactions. Such interactions are comprised of two additive terms a) stable interactions captured by using features for both users and ads whose coefficients change slowly over time, and b) ephemeral interactions that capture ad-specific residual idiosyncrasies that are missed by the stable component. Exploration is introduced via Thompson sampling on the ephemeral interactions (sample coefficients from the posterior distribution), since the stable part is estimated using large amounts of data and subject to very little statistical variance. Our model training pipeline estimates the stable part using a scatter and gather approach via the ADMM algorithm, ephemeral part is estimated more frequently by learning a per ad correction through an ad-specific logistic regression. Scoring thousands of ads at runtime under tight latency constraints is a formidable challenge when using such models, the talk will describe methods to scale such computations at runtime.Automatic Format Selection: The presentation of ads in a given slot on a page has a significant impact on how users interact with them. Web designers are adept at creating good formats to facilitate ad display but selecting the best among those automatically is a machine learning task. I will describe a machine learning approach we use to solve this problem. It is again an explore/exploit problem but the dimensionality of this problem is much less than the ad selection problem. I will also provide a detailed description of how we deal with issues like budget pacing, bid forecasting, supply forecasting and targeting. Throughout, the ML components will be illustrated with real examples from production and evaluation metrics would be reported from live tests. Offline metrics that can be useful in evaluating methods before launching them on live traffic will also be discussed.

Click-through rate (CTR) prediction is crucial in recommendation and online advertising systems. Existing methods usually model user behaviors, while ignoring the informative context which influences the user to make a click decision, e.g., click pages and pre-ranking candidates that inform inferences about user interests, leading to suboptimal performance. In this paper, we propose a Decision-Making Context Interaction Network (DCIN), which deploys a carefully designed Context Interaction Unit (CIU) to learn decision-making contexts and thus benefits CTR prediction. In addition, the relationship between different decision-making context sources is explored by the proposed Adaptive Interest Aggregation Unit (AIAU) to improve CTR prediction further. In the experiments on public and industrial datasets, DCIN significantly outperforms the state-of-the-art methods. Notably, the model has obtained the improvement of CTR+2.9%/CPM+2.1%/GMV+1.5% for online A/B testing and served the main traffic of Meituan Waimai advertising system.

In the realm of online recommendation systems, the Combinatorial Recommender (CR) system stands out for its unique approach. It presents users with a list of items on a result page, where user behavior is simultaneously influenced by contextual information and the items listed. Formulated as a combinatorial optimization problem, the objective of the CR system is to maximize the recommendation reward across the entire list of items. Despite the significant potential of CR systems, developing a practical and efficient model remains substantial challenges. These challenges stem from the dynamic nature of online environments and the pressing need for personalized recommendations. To tackle these challenges, we decompose the overarching problem into two sub-problems: list generation and list evaluation. We propose novel and pragmatic model architectures for each sub-problem aiming to concurrently enhance both effectiveness and efficiency. To further adapt the CR system to online scenarios, we integrate a bootstrap algorithm into an actor-critic reinforcement framework. This innovative approach called JD Recommender System (JDRec) is designed to continuously refine the recommendation mode through sustained user interaction, ensuring the system's adaptability and relevance. The proposed JDRec framework, tested through rigorous offline and online experiments, has shown promising results. It has been successfully deployed in online JD recommendation systems, yielding a notable improvement in click-through rate by 2.6% and augmenting the total value of the platform by 5.03%. Besides, we release the large scale dataset used in our work to facilitate further research.