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.

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.