Pseudonymized FOI Request Dataset: Korea Heritage Agency (2020–2025)

This dataset contains 573 Freedom of Information (FOI) requests submitted to Korea Heritage Agency (KHA) between January 2020 and May 2025. It supports the study "Gatekeeping AI: Predicting Vexatious Freedom of Information Requests Using Machine Learning," submitted to Public Management Review. ── Dataset Characteristics ──────────────────────────────- Records: 573 FOI requests- Time span: 2020.01.04 – 2025.05.31- Unique requesters: 148 (pseudonymized as CUST_XXXXXXXX)- Variables: 30 fields per record ── Key Findings Supported ───────────────────────────────- Vexatious request rate: 61.9% (355 of 573)- Top 10 requesters accounted for 66.1% of total volume- XGBoost classifier: Accuracy 92%, Recall 97%, AUC 0.93- 17 topics extracted via BERTopic ── Requester Type ───────────────────────────────────────- Individual (domestic): 550 (96.0%)- Organization: 15 (2.6%)- Corporation (domestic): 8 (1.4%) ── Outcome Distribution ─────────────────────────────────- Full disclosure: 119 cases (20.8%)- Partial disclosure: 5 cases (0.9%)- Non-disclosure: 3 cases (0.5%)- Terminated/vexatious: 355 cases (61.9%) · Discontinued (repeat without justification): 198 · Petition/inquiry response (non-FOI content): 157- Information not held: 61 cases (10.6%)- Withdrawn/transferred: 33 cases ── Processing ───────────────────────────────────────────- Submission method: Online (88.5%), Mobile (11.5%)- Mean processing time: 4.0 days (median: 4, max: 20) ── Pseudonymization Applied ─────────────────────────────In compliance with Korea's Personal Information Protection Act:- Requester IDs: CUST_XXXXXXXX (pre-anonymized by source system)- Staff names: STAFF_01, STAFF_02- Email addresses → [EMAIL]- Phone numbers → [TEL]- Physical addresses → [ADDR]- Personal names in free text → [NAME]- URLs → [URL] ── Variable List ────────────────────────────────────────접수번호 (Request ID), 접수일 (Date), 청구인 (Requester ID),구분 (Requester type), 청구제목 (Title), 청구내용 (Content),