model/bias/rules.py

import re

GENDER_CODED_WORDS = {
    "masculine": ["aggressive", "dominant", "competitive", "rockstar"],
    "feminine": ["supportive", "empathetic", "nurturing"]
}

PRESTIGE_SCHOOLS = {
    "iit", "nit", "mit", "stanford", "harvard", "oxford"
}

BIG_TECH = {
    "google", "amazon", "meta", "microsoft", "apple"
}


def detect_gender_coded_language(jd_text: str) -> list[str]:
    findings = []
    lower = jd_text.lower()
    for category, words in GENDER_CODED_WORDS.items():
        for w in words:
            if w in lower:
                findings.append(f"Gender-coded language detected: '{w}' ({category})")
    return findings


def detect_prestige_bias(resume_text: str) -> list[str]:
    findings = []
    lower = resume_text.lower()
    for school in PRESTIGE_SCHOOLS:
        if school in lower:
            findings.append("Prestige institution mention may influence scoring")
    return findings


def detect_company_brand_bias(resume_text: str) -> list[str]:
    findings = []
    lower = resume_text.lower()
    for company in BIG_TECH:
        if company in lower:
            findings.append("Well-known company mention may bias evaluation")
    return findings