Thompson Sampling selects a hypothesis by drawing random samples \(\theta_i \sim \text{Beta}(\alpha_i, \beta_i)\) and picking the maximum \(\theta\), whereas Upper Confidence Bound (UCB) deterministically scores hypotheses by adding an exploration bonus \(c \sqrt{\frac{\ln N}{n_i}}\) to the posterior mean \(\mu_i\).
flowchart TD
A[Available Claims] --> B{Acquisition Method}
B -->|Thompson Sampling| C["Sample θi ~ Beta(αi, βi)"]
B -->|Upper Confidence Bound| D["Score = μi + c * sqrt(ln N / ni)"]
C --> E[Select Claim with Max Value]
D --> E