ACAI Dataset — Archive of Contemporary Art Event Information (South Korea)

ACAI Dataset — Archive of Contemporary Art Event Information (South Korea) — Version:2.0.0 ACAI is a relational dataset of contemporary-art events in South Korea, built from the public websites of three national/municipal institutions — the National Museum of Modern and Contemporary Art (MMCA), the Seoul Museum of Art (SeMA) and the Asia Culture Center (ACC) — plus the national open-call portal Artnuri, from 2000 to 2025. It covers 1,775 exhibitions, 6,250 public programs (education 4,305, culture 1,945), 5,490 open calls and 17,929 collection works (MMCA 11,762, SeMA 6,167), resolved to 9,759 participants, 1,892 organizations and 698 venues, and indexed with 5,505 thematic keywords (27,073 event–keyword assignments) mapped onto 2,277 theme categories — 1,848 Getty AAT concepts, 428 local categories, and one residual category for unmatched low-frequency keywords. What's new in 2.0.0 The keyword layer was rebuilt end to end. Keywords were re-extracted from all exhibition and program descriptions under a formalized contract with verbatim text evidence and machine validation gates; surface variants were merged through deterministic normalization plus a fully logged human review. Matching to the Getty AAT was redesigned as a retrieval-then-verify procedure over the full five AAT facets (49,105 candidate concepts), decided once per keyword with an exact/close/broad/no-match verdict. Accordingly, the theme category and match quality now live on the keyword table rather than on event_keyword (schema change). In addition, 417 truncated or caption-only description texts were repaired against the live source pages, and all theme categories now carry Korean labels. Repository Contents Object Name Component Format Description ACAI_RDB_Entity.zip Relational Database CSV (9 files) 9 entity tables of the structured data representing the contemporary art events, including exhibition, program, open call, keyword and theme_category. ACAI_RDB_Relation.zip Relational Database CSV (13 files) 13 relational tables of the structured data representing the contemporary art events. ACAI_html.zip Web Archives HTML Archived source webpages of every exhibition, program and open call (13,515 gzip-compressed HTML captures). ACAI.db Relational Database (SQLite) SQLite database file Pre-built SQLite relational database containing all tables and relationships, allowing users to directly query the dataset without reconstructing the schema from CSV files. ACAI_schema.sql Database Schema SQL script SQL schema file defining the relational database structure (tables, primary keys, foreign keys, and constraints, with a comment on every column), enabling users to recreate the ACAI database structure in SQLite or other relational database systems. ACAI keyword pipeline.zip Keyword Pipeline Documentation Markdown, CSV (4 files) Documentation of the keyword extraction and matching pipeline, including the full procedure, the extraction and matching prompts given to the model, and the AAT matching decision with its justification for each of the 5,505 keywords.