In a single-slot recommendation system, users are only exposed to one item at a time, and the system cannot collect user feedback on multiple items simultaneously. Therefore, only pointwise modeling solutions can be adopted, focusing solely on modeling the likelihood of clicks or conversions for items by users to learn user-item preferences, without the ability to capture the ranking information among different items directly. However, since user-side information is often much more abundant than item-side information, the model can quickly learn the differences in user intrinsic tendencies, which are independent of the items they are exposed to. This can cause these intrinsic tendencies to become a shortcut bias for the model, leading to insufficient mining of the most concerned user-item preferences. To solve this challenge, we introduce the Breaker model. Breaker integrates an auxiliary task of user representation clustering with a multi-tower structure for cluster-specific preference modeling. By clustering user representations, we ensure that users within each cluster exhibit similar characteristics, which increases the complexity of the pointwise recommendation task on the user side. This forces the multi-tower structure with cluster-driven parameter learning to better model user-item preferences, ultimately eliminating shortcut biases related to user intrinsic tendencies. In terms of training, we propose a delayed parameter update mechanism to enhance training stability and convergence, enabling end-to-end joint training of the auxiliary clustering and classification tasks. Both offline and online experiments demonstrate that our method surpasses the baselines. It has already been deployed and is actively serving tens of millions of users daily on Meituan, one of the most popular e-commerce platforms for services.

Phone number recycling (PNR) refers to the event wherein a mobile operator collects a disconnected number and reassigns it to a new owner. It has posed a threat to the reliability of the existing authentication solution for e-commerce platforms. Specifically, a new owner of a reassigned number can access the application account with which the number is associated, and may perform fraudulent activities. Existing solutions that employ a reassigned number database from mobile operators are costly for e-commerce platforms with large-scale users. Thus, alternative solutions that depend on only the information of the applications are imperative. In this work, we study the problem of detecting accounts that have been compromised owing to the reassignment of phone numbers. Our analysis on Meituan’s real-world dataset shows that compromised accounts have unique statistical features and temporal patterns. Based on the observations, we propose a novel model called temporal pattern and statistical feature fusion model (TSF) to tackle the problem, which integrates a temporal pattern encoder and a statistical feature encoder to capture behavioral evolutionary interaction and significant operation features. Extensive experiments on the Meituan and IEEE-CIS datasets show that TSF significantly outperforms the baselines, demonstrating its effectiveness in detecting compromised accounts due to reassigned numbers.

With the explosive development of e-commerce for service, tens of millions of orders are generated every day on the Meituan platform. By allocating bonuses to new customers when they pay, the Meituan platform encourages them to use its own payment service for a better experience in the future. It can be formulated as a multi-choice knapsack problem (MCKP), and the mainstream solution is usually a two-stage method. The first stage is user intent detection, predicting the effect for each bonus treatment. Then, it serves as the objective of the MCKP, and the problem is solved in the second stage to obtain the optimal allocation strategy. However, this solution usually faces the following challenges: (1) In the user intent detection stage, due to the sparsity of interaction and noise, the traditional multi-treatment effect estimation methods lack interpretability, which may violate the domain knowledge that the marginal gain is non-negative with the increase of the bonus amount in economic theory. (2) There is an optimality gap between the two stages, which limits the upper bound of the optimal value obtained in the second stage. (3) Due to changes in the distribution of orders online, the actual cost consumption often violates the given budget limit. To solve the above challenges, we propose a framework that consists of three modules, i.e., User Intent Detection Module, Online Allocation Module, and Feedback Control Module. In the User Intent Detection Module, we implicitly model the treatment increment based on deep representation learning and constrain it to be non-negative to achieve monotonicity constraints. Then, in order to reduce the optimality gap, we further propose a convex constrained model to increase the upper bound of the optimal value. For the third challenge, to cope with the fluctuation of online bonus consumption, we leverage a feedback control strategy in the framework to make the actual cost more accurately approach the given budget limit. Finally, we conduct extensive offline and online experiments, demonstrating the superiority of our proposed framework, which reduced customer acquisition costs by 5.07% and is still running online.

Online-to-Offline (O2O) e-commerce service platforms and their users are faced with various fraud risks. Among them, financial identity theft is a widely existing challenge. However, existing methods are insufficient to detect this type of fraud. In this paper, we address the financial identity theft detection problem in e-commerce services by leveraging access environment and behavior sequence. To explore the fraud patterns, we first make a detailed analysis using real cases of identity theft from Meituan, a leading O2O e-commerce platform in China. Our findings are twofold. First, fraudulent accounts sharing the same personal ID would have different access environments, such as devices and IP addresses. Second, a group of fraudulent accounts may have aggregations of devices, IP addresses, and delivery addresses. Based on these observations, we propose a hybrid method termed EnvIT to detect financial identity theft based on the heterogeneous graph and the behavior sequence. EnvIT is able to characterize the access environment and the historical behavior of the accounts. Furthermore, an attentive module is adopted to assign weights to different features automatically. We further evaluate EnvIT via extensive experiments using a real-world dataset from Meituan. Our experimental results demonstrate that EnvIt outperforms several baseline methods in fraudulent account detection and achieves an AUC of 0.9210.

In most real-world large-scale online applications (e.g., e-commerce or finance), customer acquisition is usually a multi-step conversion process of audiences. For example, an impression->click->purchase process is usually performed of audiences for e-commerce platforms. However, it is more difficult to acquire customers in financial advertising (e.g., credit card advertising) than in traditional advertising. On the one hand, the audience multi-step conversion path is longer, an impression->click->application->approval->activation process usually occurs during the audience conversion for credit card business in financial advertising. On the other hand, the positive feedback is sparser (class imbalance) step by step, and it is difficult to obtain the final positive feedback due to the delayed feedback of activation. Therefore, it is necessary to use the positive feedback information of the former step to alleviate the class imbalance of the latter step. Multi-task learning is a typical solution in this direction. While considerable multi-task efforts have been made in this direction, a long-standing challenge is how to explicitly model the long-path sequential dependence among audience multi-step conversions for improving the end-to-end conversion. In this paper, we propose an Adaptive Information Transfer Multi-task (AITM) framework, which models the sequential dependence among audience multi-step conversions via the Adaptive Information Transfer (AIT) module. The AIT module can adaptively learn what and how much information to transfer for different conversion stages. Besides, by combining the Behavioral Expectation Calibrator in the loss function, the AITM framework can yield more accurate end-to-end conversion identification. The proposed framework is deployed in Meituan app, which utilizes it to real-timely show a banner to the audience with a high end-to-end conversion rate for Meituan Co-Branded Credit Cards. Offline experimental results on both industrial and public real-world datasets clearly demonstrate that the proposed framework achieves significantly better performance compared with state-of-the-art baselines. Besides, online experiments also demonstrate significant improvement compared with existing online models. Furthermore,we have released the source code of the proposed framework at https://github.com/xidongbo/AITM.

The Internet was designed with the end-to-end principle where the network layer provided merely the best-effort forwarding service. This design makes it challenging to add new services into the Internet infrastructure. However, as the Internet connectivity becomes a commodity, users and applications increasingly demand new in-network services. This paper proposes PacketCloud, a cloudlet-based open platform to host in-network services. Different from standalone, specialized middleboxes, cloudlets can efficiently share a set of commodity servers among different services, and serve the network traffic in an elastic way. PacketCloud can help both Internet Service Providers (ISPs) and emerging application/content providers deploy their services at strategic network locations. We have implemented a proof-of-concept prototype of PacketCloud. PacketCloud introduces a small additional delay, and can scale well to handle high-throughput data traffic. We have evaluated PacketCloud in both a fully functional emulated environment, and the real Internet.