Cross-domain sequential recommendation (CDSR) tackles data sparsity and cold-start issues by leveraging information from the source domain to enhance prediction accuracy in the target domain. However, the recommendation fairness issue may further deteriorate dramatically with the biased knowledge transfer of overlapped users. This paper is the first study to address and improve fairness measurement between different demographic groups in CDSR. The proposed FairCDSR employs sequence augmentation techniques to enrich the interaction histories of disadvantaged user groups, which typically have less training data. These augmented sequences are further represented by a contrastive learning method with hard negative sampling to mitigate the unfairness in recommendations. Then, to more precisely capture cross-domain preferences, a multi-interest learning approach is applied to each group across the domains. Finally, an interest-level knowledge transfer algorithm with fixed bandwidth limitations for each group is developed to extract fair and semantic cross-domain information. Extensive experiments conducted on real-world datasets demonstrate the effectiveness of FairCDSR. Compared to existing cross-domain or fair recommendation systems, FairCDSR significantly reduces recommendation disparity between advantaged and disadvantaged groups. Benefiting from a significant improvement in the recommendation accuracy of the disadvantaged group, the overall system performance can also be effectively enhanced by 5-10%.

The significant increase in credit card transactions can be attributed to the rapid growth of online shopping and digital payments, particularly during the COVID-19 pandemic. To safeguard cardholders, e-commerce companies, and financial institutions, the implementation of an effective and real-time fraud detection method using modern artificial intelligence techniques is imperative. However, the development of machine-learning-based approaches for fraud detection faces challenges such as inadequate transaction representation, noise labels, and data imbalance. Additionally, practical considerations like dynamic thresholds, concept drift, and verification latency need to be appropriately addressed. In this study, we designed a fraud detection method that accurately extracts a series of spatial and temporal representative features to precisely describe credit card transactions. Furthermore, several auxiliary self-supervised objectives were developed to model cardholders’ behavior sequences. By employing intelligent sampling strategies, potential noise labels were eliminated, thereby reducing the level of data imbalance. The developed method encompasses various innovative functions that cater to practical usage requirements. We applied this method to two real-world datasets, and the results indicated a higher F1 score compared to the most commonly used online fraud detection methods.

The growing number of publications in the field of artificial intelligence highlights the need for researchers to enhance their efficiency in searching for relevant articles. Most paper recommendation models either rely on simplistic citation relationships among papers or focus on content-based approaches, both of which overlook interactions within academic networks. To address the aforementioned problem, knowledge graph embedding (KGE) methods have been used for citation recommendations because recent research proves that graph representations can effectively improve recommendation model accuracy. However, academic networks are dynamic, leading to changes in the representations of users and items over time. The majority of KGE-based citation recommendations are primarily designed for static graphs, thus failing to capture the evolution of dynamic knowledge graph (DKG) structures. To address these challenges, we introduced the evolving knowledge graph embedding (EKGE) method. In this methodology, evolving knowledge graphs are input into time-series models to learn the patterns of structural evolution. The model has the capability to generate embeddings for each entity at various time points, thereby overcoming limitation of static models that require retraining to acquire embeddings at each specific time point. To enhance the efficiency of feature extraction, we employed a multiple attention strategy. This helped the model find recommendation lists that are closely related to a user’s needs, leading to improved recommendation accuracy. Various experiments conducted on a citation recommendation dataset revealed that the EKGE model exhibits a 1.13% increase in prediction accuracy compared to other KGE methods. Moreover, the model’s accuracy can be further increased by an additional 0.84% through the incorporation of an attention mechanism.

Graph-based cross-domain recommendations (CDRs) are useful for suggesting appropriate items because of their promising ability to extract features from user–item interactions and transfer knowledge across domains. Thus, the model can effectively alleviate cold start and data sparsity issues. Although the graph-based CDRs can capture valuable information, they still have some limitations. First, embeddings are highly vulnerable to noisy interactions, because the message aggregation in the graph convolutional network can further enlarge the impact. Second, because of the property of graph-structured data, the influence of high-degree nodes on representation learning is more than that of the long-tail items, and this can cause a poor recommendation performance. In this study, we devised a novel A daptive A dversarial C ontrastive L earning framework for graph-based C ross- D omain R ecommendation ( ACLCDR ). The ACLCDR introduces reinforcement learning to generate adaptive augmented samples for contrastive learning tasks. Then, we leveraged a multitask training strategy to jointly optimize the model with auxiliary tasks. Finally, we verified the effectiveness of the ACLCDR through nine real-world cross-domain tasks adopted from Amazon and Douban. We observed that ACLCDR exceeded the best state-of-the-art baseline by 25%, 42.5%, 16.3%, and 23.8% in terms of HR@ 10 and NDCG@10 for the Music & Movie task from Amazon.

The data of recommendation systems typically only contain the purchased item as positive data and other un-purchased items as unlabeled data. To train a good recommendation model, in addition to the known positive information, we also need high-quality negative information. Capturing negative signals in positive and unlabeled data is challenging for recommendation systems. Most studies have used specific data and proposed negative sampling methods suitable to the data characteristics. Existing negative sampling strategies cannot automatically select suitable approaches for different data. However, this one-size-fits-all strategy often makes potential positive samples considered as negative, or truly negative samples considered as potential positive samples and recommend to users. In this way, it will not only turn down the recommendation result, but even also have an adverse effect. Accordingly, we propose a novel negative sampling model, Reinforced PU-learning with Hybrid Negative Sampling Strategies for Recommendation (RHNSR), which can combine multiple sampling strategies and dynamically adjust the proportions used by different sampling strategies. In addition, ensemble learning, which integrates various model sampling strategies for obtaining an improved solution, was applied to RHNSR. Extensive experiments were conducted on three real-world recommendation datasets, and the experimental results indicated that the proposed model significantly outperformed state-of-the-art baseline models and revealed significant improvements in precision and hit ratio (49.02% and 37.41%, respectively).

Next-item recommendation involves predicting the next item of interest of a given user from their past behavior. Users tend to browse and purchase various items on e-commerce websites according to their varied interests and needs, as reflected in their purchasing history. Most existing next-item recommendation methods aim at extracting the main point of interest in each browsing session and encapsulate it in a single representation. However, past behavior sequences reflect the multiple interests of a single user, which cannot be captured by methods that focus on single-interest contexts. Indeed, multiple interests cannot be captured in a single representation, and doing so results in missing information. Therefore, we propose a model with a multi-interest structure for capturing the various interests of users from their behavior sequence. Moreover, we adopted a method based on a graph neural network to construct interest graphs based on the historical and current behavior sequences of users. These graphs can capture complex item transition patterns related to different interests. In experiments, the proposed method outperforms state-of-the-art session-based recommendation systems on three real-world datasets, achieving 4% improvement of Recall over the SOTAs on Jdata dataset.

Behavioral finance is a relatively new and developing research field which adopts cognitive psychology and emotional bias to explain the inefficient market phenomenon and some irrational trading decisions. Unlike the experts in this field who tried to reason the price anomaly and applied empirical evidence in many different financial markets, we employ the advanced binary classification algorithms, such as AdaBoost and support vector machines, to precisely model the overreaction and strengthen the portfolio compositions of the contrarian trading strategies. The novelty of this article is to discover the financial time-series patterns through a high-dimensional and nonlinear model which is constructed by integrated knowledge of finance and machine learning techniques. We propose a dual-classifier learning framework to select candidate stocks from the past results of original contrarian trading strategies based on the defined learning targets. Three different feature extraction methods, including wavelet transformation, historical return distribution, and various technical indicators, are employed to represent these learning samples in a 381-dimensional financial time-series feature space. Finally, we construct the classifier models with four different learning kernels and prove that the proposed methods could improve the returns dramatically, such as the 3-year return that improved from 26.79% to 53.75%. The experiments also demonstrate significantly higher portfolio selection accuracy, improved from 57.47% to 66.41%, than the original contrarian trading strategy. To sum up, all these experiments show that the proposed method could be extended to an effective trading system in the historical stock prices of the leading U.S. companies of S&P 100 index.