match-score

name	match-score
description	Compute a quantitative match score between a resume and job requirements. Evaluates per-category alignment (must_have, nice_to_have, culture, logistics), identifies strengths and gaps, and produces an overall weighted score from 0 to 100. Use when assessing how well a resume matches a specific job description.

Match Score

You are tasked with evaluating how well a resume matches a set of structured job requirements.

Input

You receive two inputs:

1. resumeData: The resume content (either structured ResumeData JSON or raw .tex source containing the resume content)
2. jobRequirements: A structured JobRequirements JSON object with categorized requirements, keywords, and importance weights

Output

Return a JSON object matching this schema:

{
  "overallScore": 72,
  "categoryScores": {
    "must_have": 68,
    "nice_to_have": 80,
    "culture": 75,
    "logistics": 90
  },
  "matches": [
    {
      "requirementId": "req_1",
      "evidence": "Resume mentions 5 years of Python in Experience section",
      "strength": "strong"
    }
  ],
  "gaps": [
    {
      "requirementId": "req_3",
      "description": "No mention of Kubernetes experience",
      "severity": "high",
      "bridgeable": true,
      "bridgeHint": "Docker experience could be reframed to show container orchestration familiarity"
    }
  ],
  "keywordCoverage": {
    "found": ["Python", "AWS", "Docker"],
    "missing": ["Kubernetes", "GraphQL"]
  }
}

Deduplication Rules (Strict)

These rules are mandatory. Violating them produces a useless report.

1. One entry per requirement

Each requirementId must appear in exactly one entry across the entire output — either in matches OR in gaps, never in both. A requirement is either covered or it isn't.

2. Distinct evidence within matches

Each entry in matches must describe a different skill or capability matched by the resume. Do not create multiple entries that express the same strength in different words.

• ❌ Bad: separate entries for "5 years Python" and "Python used in ML project" both pointing to req_1
• ✅ Good: one entry per requirement, summarising the strongest single piece of evidence

3. Distinct missing capabilities in gaps

Each entry in gaps must describe a separate, concrete missing capability. Consolidate related gaps into one entry rather than listing synonyms or closely related items individually.

• ❌ Bad: separate gaps for "no Kubernetes" and "no container orchestration" — these are the same concept
• ✅ Good: one gap entry per distinct skill/domain that the resume does not address

4. Prioritisation

• matches: return only the top strongest matches (up to 5), ordered by strength desc (strong → moderate → weak)
• gaps: return all significant gaps, ordered by severity desc (high → medium → low); omit low-severity gaps if there are already 8+ entries

Scoring Algorithm

Per-Category Scoring (0–100)

For each category, compute a weighted average:

categoryScore = sum(requirementScore_i * importance_i) / sum(importance_i)

Where requirementScore_i is:

• 100: Strong match — direct evidence in resume
• 70: Partial match — related experience that could bridge the gap
• 30: Weak match — tangentially related content
• 0: No match — no evidence found

Overall Score

overallScore = must_have * 0.50 + nice_to_have * 0.20 + culture * 0.15 + logistics * 0.15

The weights reflect that must_have requirements are the primary hiring signal.

Match Strength Classification

• strong: Direct, explicit evidence (e.g., "5 years Python" for "Python required")
• moderate: Related but not exact (e.g., "Java experience" for "Python required")
• weak: Tangential connection only

Gap Severity Classification

• high: Missing a must_have requirement with importance > 0.7
• medium: Missing a must_have requirement with importance <= 0.7, or a nice_to_have with importance > 0.6
• low: Missing a nice_to_have or culture/logistics item

Bridgeability Assessment

For each gap, assess whether the candidate's existing experience could be reframed to partially address it:

• bridgeable: true — The candidate has transferable skills (e.g., has Docker experience for a Kubernetes requirement)
• bridgeable: false — The requirement cannot be addressed with existing experience (e.g., needs specific certification the candidate lacks)
• Include a bridgeHint for bridgeable gaps explaining the potential reframe

Rules

• Score objectively based on evidence, not assumptions
• Don't inflate scores — be realistic about gaps
• Consider both explicit skills and implied experience from job descriptions
• Keywords appearing in context (bullet points) are stronger signals than appearing in a skills list alone

For detailed scoring rubric and calibration examples, consult references/RUBRIC.md.