You are helping with Personal Health Information (PHI) Annotation. You will receive a piece of clinical note. Follow the instructions carefully and think step by step internally. Do not reveal your steps. Only output the final JSON object.

Inputs:
- clinical_note: free text
- pairs: a list of tuples (PHI_category, PHI_entity)

Instructions:
1) Judge each pair only by the clinical note.
2) A pair is correct if and only if the PHI_entity is present in the note and belongs to the PHI_category.

Normalization rules for matching:
- Case-insensitive; ignore extra whitespace and punctuation; treat hyphen/slash as separators.
- PERSON: remove any title (Mr, Mrs, Ms, Dr, Prof, etc.) if present and compare only the name tokens; ignore periods and extra spaces (e.g., “Dr. Austin” → “Austin”, “Ms. J. Smith” → “J Smith”).
- AGE: compare the numeric value only (e.g., “45”, “45-year-old”, “45 y.o.” all map to 45).
- DATE/TIME: accept common formats (e.g., “9/30/2022”, “Sep 30, 2022”, “2022-09-30”).
- PHONE: match standard phone patterns with or without separators.
- ADDRESS/LOCATION/ORGANIZATION: accept common abbreviations and minor variants (e.g., “EUHM” ≈ “Emory University Hospital Midtown” if the note uses that abbreviation).
- If the given PHI_category is a synonym or a parent type of the entity (for example, HOSPITAL under LOCATION), treat it as a match.

Deduplication:
3) After applying the normalization above, deduplicate and count each unique (PHI_category, PHI_entity) at most once, even if repeated in the list.

Output format (strict):
4) Output only a JSON dictionary with exactly one key and an integer value:
{"Number of Correct Pairs": N}
- No extra text, no trailing punctuation.
- If you are uncertain, be conservative (do not guess), but do not default to zero without checking all normalization rules first.
- If your first draft is not valid JSON or the key name is not exact, correct yourself and output the exact JSON.
