domain-anchoring

Domain Anchoring (Phase 0)

Overview

Bind the correct domain identity before any research work begins. Wrong expertise produces irrelevant baselines, missed evaluation criteria, and reviewer rejection.

When to Use

digraph trigger {
    "User mentions research?" [shape=diamond];
    "research-anchor.yaml exists?" [shape=diamond];
    "Confirmed by user?" [shape=diamond];
    "Invoke domain-anchoring" [shape=box];
    "Skip — already anchored" [shape=box];

    "User mentions research?" -> "research-anchor.yaml exists?" [label="yes"];
    "User mentions research?" -> "Invoke domain-anchoring" [label="no → not research"];
    "research-anchor.yaml exists?" -> "Confirmed by user?" [label="yes"];
    "research-anchor.yaml exists?" -> "Invoke domain-anchoring" [label="no"];
    "Confirmed by user?" -> "Skip — already anchored" [label="yes"];
    "Confirmed by user?" -> "Invoke domain-anchoring" [label="no"];
}

Step 1: Identify Domain and Subdomain

Extract the research domain (ML, bioinformatics, physics, chemistry, multimedia, data mining, etc.) and subdomain (e.g., few-shot learning, single-cell genomics, molecular dynamics). If ambiguous, ask the user directly.

Step 2: Classify Research Type

Determine one of four types:

Type	Focus	Typical Venue
M — Method/Model	New algorithm or architecture	ML conferences (NeurIPS, ICML, ICLR)
D — Discovery/Data	Data-driven scientific insights	Domain journals (Nature Methods, Cell)
C — Computational Tool	Pipelines, software, infrastructure	Software journals (JOSS, Bioinformatics)
H — Hybrid	New method + domain application	Cross-disciplinary venues

If type is unclear, MUST ask clarifying questions. Do not guess.

Step 3: Anchor Expert Persona

Bind the correct domain expert identity. Use this table to set reviewer_focus in the anchor file: