find-annotate-dupes.py

import csv

INCLUDED_FIELDS = (
    'classification_id',
    'subject_ids',
    'user_name',
    'metadata',
    'annotations'
)

with open('annotate-classifications.csv') as f:
    r = csv.DictReader(f)
    user_subjects = {}
    for row in r:
        user_subjects.setdefault(
            row['user_name'],
            {}
        ).setdefault(
            row['subject_ids'],
            []
        ).append(tuple(
            row[field] for field in INCLUDED_FIELDS
        ))

with open('duplicate-annotate-classifications.csv', 'w') as out_f:
    w = csv.writer(out_f)
    w.writerow(INCLUDED_FIELDS)
    for user, subjects_dict in user_subjects.items():
        for subject_id, rows in subjects_dict.items():
            if len(rows) > 1:
                for row in rows:
                    w.writerow(row)