In the user growth scenario, Internet companies invest heavily in paid acquisition channels to acquire new users. But sustainable growth depends on acquired users' generating lifetime value (LTV) exceeding customer acquisition cost (CAC). In order to maximize LTV/CAC ratio, it is crucial to predict channel-level LTV in an early stage for further optimization of budget allocation. The LTV forecasting problem is significantly different from traditional time series forecasting problems, and there are three main challenges. Firstly, it is an unaligned multi-time series forecasting problem that each channel has a number of LTV series of different activation dates. Secondly, to predict in the early stage, it faces the imbalanced short-input long-output (SILO) challenge. Moreover, compared with the commonly used time series datasets, the real LTV series are volatile and non-stationary, with more frequent fluctuations and higher variance. In this work, we propose a novel framework called Trapezoidal Temporal Fusion (TTF) to address the above challenges. We introduce a trapezoidal multi-time series module to deal with data unalignment and SILO challenges, and output accurate predictions with a multi-tower structure called MT-FusionNet. The framework has been deployed to the online system for Douyin. Compared to the previously deployed online model, MAPE_p decreased by 4.3%, and MAPE_a decreased by 3.2%, where MAPE_p denotes the point-wise MAPE of the LTV curve and MAPE_a denotes the MAPE of the aggregated LTV.

Bike sharing systems have been widely deployed around the world in recent years. A core problem in such systems is to reposition the bikes so that the distribution of bike supply is reshaped to better match the dynamic bike demand. When the bike-sharing company or platform is able to predict the revenue of each reposition task based on historic data, an additional constraint is to cap the payment for each task below its predicted revenue. In this paper, we propose an incentive mechanism called TruPreTar to incentivize users to park bicycles at locations desired by the platform toward rebalancing supply and demand. TruPreTar possesses four important economic and computational properties such as truthfulness and budget feasibility. Furthermore, we prove that even when the payment budget is tight, the total revenue still exceeds or equals the budget. Otherwise, TruPreTar achieves 2-approximation as compared to the optimal (revenue-maximizing) solution, which is close to the lower bound of at least √2 that we also prove. Using an industrial dataset obtained from a large bike-sharing company, our experiments show that TruPreTar is effective in rebalancing bike supply and demand and, as a result, generates high revenue that outperforms several benchmark mechanisms.

Uber, Lyft and some other companies achieve great business success in the area of ridesharing. However, there are still many issues to be addressed, for instance, the lack of a well-designed rush hour (i.e., when supply is much less than demand) pricing strategy, the seat resources waste and so on. The online ridesharing model fits better to reality but brings significant design challenges. In this paper, we propose a complementary online auction design: No Preemption auction. The auction in ridesharing needs to deal with dynamic costs, which is different from most of the existing auctions and brings significant challenges. We design corresponding online auction mechanism for this model. Our mechanism is truthful with Risk Aversion Hypothesis, budget balanced (i.e., the payment is not less than the cost), computational efficient, individually rational, and make the best use of the seat resources. We give theoretical proof of the truthfulness, budget balance, computational efficiency and individual rationality. We also give a discussion about the general online ridesharing mechanism performance. Finally, we evaluate the performance of our mechanism based on real taxi-trace data in Shanghai. The results show that our mechanism can achieve rather good revenue and social welfare.

With the development of 60 GHz technology, data centers are going to be wireless. A fundamental challenge in wireless data center networking is how to efficiently use 60 GHz wireless technology to improve the performance. Many existing works have been proposed for this, but most of them do not perform well in connectivity or may not be flexible for different environments in data centers. This paper presents Graphite, a novel network structure that has many desirable features for wireless data centers. The whole architecture can be suitable for several different deployments of data centers. In Graphite, the problem of link blockage can be properly solved. Graphite makes best use of the propagation distance of 60 GHz and allows one server to communicate with as many other servers as possible. Graphite also improves the average node degree, which is more than any other existing wireless topologies on the same condition. Furthermore, Graphite can be suitable for data center with different deployments. We build a small testbed of Graphite to demonstrate its ability to solve the problem of link blockage. Results from theoretical analysis and extensive evaluations show that Graphite is a viable wireless topology for data center networks.