A reputation-based incentive and allocation model using double auction for mobile crowdsensing

A Trust Incentive-Driven Consensus Feedback Mechanism for Large-Scale Participatory Budgeting

Large-scale participatory budgeting (PB) lacks a mechanism for reconciling divergent resident preferences before project selection. Conventional consensus feedback mechanisms are difficult to deploy in this setting because anonymous ballots provide no direct basis for tailoring feedback, while resident-by-resident adjustment is computationally infeasible at population scale. We therefore develop a trust incentive-driven consensus feedback mechanism that transforms ballots into structural and behavioral signals. We apply fuzzy c-means (FCM) to identify overlapping preference groups and use the resulting memberships to construct group-mediated proxy trust, while using supported project costs to infer residents’ risk attitudes. These signals are incorporated into an extended interactive trust function (E-ITF) and a bounded consensus model that generates differentiated trust incentives and resident-specific adjustment bounds. Then, we develop the trust incentive-driven bounded consensus (TIBC) algorithm, which identifies selected inconsistent residents as a batch, determines bounded adjustment coefficients, and iteratively updates preferences, proxy trust, and resident weights. Finally, we evaluate the proposed mechanism using a real municipal PB case. TIBC algorithm reaches the common consensus target in two iterations, while maintaining subgroup consensus balance. These findings provide decision makers with a transparent and scalable way to incorporate consensus feedback into anonymous PB without requiring observed interpersonal trust data.

Two-Stage Auctions with Bid Refinement for Online Advertising

To balance prediction accuracy and system latency, large-scale online ad auctions employ two-stage architectures. These systems first retrieve a candidate ads subset using coarse quality metrics before finalizing auction outcomes with refined metrics. However, existing implementations typically elicit user-specific bids only once, overlooking the impact of real-time quality metrics on advertiser valuations and thereby limiting allocation efficiency. Motivated by recent industry practice, we investigate the design of two-stage auctions that allow advertisers to submit and update their bids, with second-stage bids serving as refinements of the initial ones. We derive the incentive-compatible (IC) conditions and analyze the revenue properties of this two-stage auction. Notably, an additional entry fee is required to prevent inflated initial bids, which compromises the standard ex-post individual rationality (IR) property. To address this, we propose a dynamic two-stage auction that adopts realization-dependent entry fees with discounts. By leveraging historical bidding information, our mechanism guarantees approximate ex-ante IC across both stages and restores ex-post IR.

Incentive-Compatible Diffusion Combinatorial Auctions

Learning to Allocate Incentives for Incentivized Advertising via Offline Model-Based Reinforcement Learning

An offline model-based RL framework for cost-controllable sequential incentive allocation is developed and an independent counterfactual scorer evaluates each learned policy on held-out logs, enabling pre-launch selection without costly online exposure.

Promotion Thresholds, Revenue Sharing, and Delivery Risk in Reward-Based Crowdfunding

This paper investigates two marketing strategies a reward-based crowdfunding platform employs to align its preferences with an entrepreneur’s choice of pledge and target levels. These are (a) how to promote campaigns to potential backers, and (b) how to share campaign revenues with the entrepreneur. Kickstarter, for instance, promotes a set of campaigns by compiling a list of “recommended” projects. This research shows that the platform’s choice of the promotion rule may expose entrepreneurs to the risk of not generating sufficient funds to start production, which can damage their and the platform’s reputation. When the platform’s reputational risk is not very high, it reduces the risk of non-delivery by increasing the revenue share of the entrepreneur. The platform’s strategies are likely to ensure production when backers derive warm glow from pledging, when the entrepreneur’s development cost is low, or when the entrepreneur has minimal reputational cost if production fails. However, low reputational costs motivate the entrepreneur to lower the target, thus increasing the likelihood of insufficient funds to start production. We propose strategies the platform can use, including customizing the revenue share based on the campaign characteristics, to rectify such misalignments.

Beyond Isolated Bidding: Campaign-Channel Coordination for Global Optimal Returns in Online Advertising

Automated bidding is a core component of modern online advertising systems. With the rapid proliferation of heterogeneous media channels, bidding strategies are required to handle ad requests across multiple channels in a unified manner to maximize total conversions while satisfying shared budget and cost-per-acquisition (CPA) constraints. Unlike existing approaches that decompose budget and CPA constraints into individual channels, we propose Campaign–Channel Coordinated Bidding (C3Bid), a bias-aware automated bidding framework that jointly optimizes final bidding decisions by coordinating campaign and channels under a unified campaign-level objective and constraints. The key insight of C3Bid is to leverage channel-wise estimation bias as an explicit coupling mechanism, where campaign-level and channel-level bids are combined through bias-aware weighting to produce the final bid. We theoretically show that the unified formulation admits a larger feasible solution space and achieves optimal expected conversions under shared constraints, while the proposed bias-aware coordination mechanism further improves channel scalability and robustness to cold-start scenarios. Extensive offline experiments and large-scale online A/B tests in a real-world advertising system demonstrate consistent and significant gains in conversions, budget utilization, and bidding stability under CPA constraints.