Evaluating Inter-Column Logical Relationships in Synthetic Tabular Data Generation

Yunbo Long, Liming Xu, Alexandra Brintrup · 2025

Read Evaluating Inter-Column Logical Relationships in Synthetic Tabular Data Generation by Yunbo Long, Liming Xu, Alexandra Brintrup (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-04055.txt.

Description

Yunbo Long, Liming Xu, and Alexandra Brintrup built tests for whether synthetic data tables keep their logic straight, in a paper published in 2025. Synthetic tables guard privacy and feed model training. Standard grades check how well the spread of values matches the real table. They skip whether the columns still obey real-world logic: rank orders in a firm, place nested in place, or time steps in line.

Their three scores grade just that preservation of logic. On a real trade dataset, both old and new table generators often broke the logic. Rank and time links failed most. The authors map paths to fix it, by building logic into the generation step. Code is free. They say logic grades should join spread grades in each synthetic table report.

Catalog record

CREATORYunbo Long, Liming Xu, Alexandra Brintrup
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE31 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-04055

Browse the full library · About the Tower of Babel