Add biomedical domain preset (PubMed / BioASQ)

[vignesh2027]

Task

Add a biomedical domain preset calibrated for PubMed-style literature retrieval.

Proposed values:

• τ = 0.32 (between scientific 0.30 and medical 0.35)
• Higher γ (causal weight) — biomedical causality is dense
• δ_SDC = 0.74

Files to change:

• core/sdc.py — add biomedical key to DOMAIN_PRESETS
• tests/test_sdc.py — add preset test

Comment below to get assigned — great first contribution!

[vignesh2027]

Implemented and merged in commit 8f5ddff!

Change in core/sdc.py:

DOMAIN_TAUS = {
    "scientific":  0.30,   # strictest
    "biomedical":  0.32,   # NEW — PubMed/BioASQ, sits between scientific and medical
    "medical":     0.35,
    ...
}

Why τ = 0.32: Biomedical literature has very dense causal chains (gene → protein → pathway → disease), stricter than general medical (0.35) but slightly more lenient than pure scientific (0.30) which requires exact progenitor chain matching.

Use it:

from vortexrag import VortexRAG, VortexRAGConfig

rag = VortexRAG(corpus=pubmed_abstracts, config=VortexRAGConfig(domain="biomedical"))
rag.index()
result = rag.query("What is the role of TNF-alpha in sepsis?")

6 new tests added in tests/test_sdc.py — all passing.

If you'd like to calibrate this further on BioASQ or PubMedQA benchmarks, a PR with updated τ values and supporting benchmark numbers would be very welcome!