The acquisition policy employs a two-stage selection mechanism that first samples a Rank Factor Class weighted by class freshness, then selects specific claims within that class using Thompson sampling, UCB, or epsilon-greedy selection.

flowchart TD
    A[All Active Claims] --> B[Stage 1: Freshness Weighting]
    B --> C[Sample Rank Factor Class]
    C --> D[Stage 2: Bayesian Strategy]
    D --> E[Thompson / UCB / eps-Greedy]
    E --> F[Selected Claim for Ideation]