The taxonomy is a locked two-level hierarchy comprising seven core classes of causal mechanisms and 33 seeded sub-level rank factors that categorize how content features influence LLM ranking preferences, expandable via an evidence-driven custodian flow.
flowchart TD
Taxonomy["Rank Factor Taxonomy"] --> C1["Parametric Recall"]
Taxonomy --> C2["Grounding & Citations"]
Taxonomy --> C3["Relevance & Semantics"]
Taxonomy --> C4["Authority & Trust"]
Taxonomy --> C5["Structure & Layout"]
Taxonomy --> C6["Temporal Freshness"]
Taxonomy --> C7["Intent Alignment"]
C1 --> F1["Entity Association"]
C2 --> F2["Source Verifiability"]
C3 --> F3["Latent Topic Overlap"]
C4 --> F4["Domain Prominence"]
C5 --> F5["Snippet Extractability"]
C6 --> F6["Fact Recency"]
C7 --> F7["Query Task Fit"]Referenced by