The automated classifier executes a structured, post-seeding LLM pass to tag each generated claim with positional RFID codes from the active taxonomy, enabling freshness-based acquisition weighting without exposing the taxonomy directly to the hypothesis generator.

flowchart LR
    A[Seeder: Blind to Taxonomy] -->|Raw Hypotheses| B[Automated Classifier]
    Tax[Locked Active Taxonomy] -->|Taxonomy Codes / RFIDs| B
    B -->|Tagged Claims| C[Bayesian Acquisition Engine]
    C -->|Two-Stage Selection| D[Ideator Edit Loop]