Logo image
KnowGen: Modeling Synthetic Dataset Construction as Inference over Latent Acceptability
Conference paper   Open access

KnowGen: Modeling Synthetic Dataset Construction as Inference over Latent Acceptability

Moumita Chanda, Alexander Maas and Hasan M Jamil
Workshop on Synthetic Data Generation and Management for Building AI Systems, pp.46-54
ACM Conferences, ACM
SIGMOD/PODS '26: International Conference on Management of Data (Bengaluru, India, 05/31/2026–06/05/2026)
05/31/2026

Abstract

Applied computing -- Life and medical sciences Computing methodologies -- Computer vision Computing methodologies -- Knowledge representation and reasoning Computing methodologies -- Machine learning approaches
Synthetic data are increasingly used in medical imaging and other data-scarce domains, yet existing approaches focus primarily on realism or downstream utility, leaving the problem of dataset validation underexplored. This paper introduces KnowGen, a framework that models synthetic dataset construction as inference over latent acceptability, defined as the fitness-for-use of generated instances under structured specifications. KnowGen treats synthetic generation as a specification-driven process and maps each instance to structured evidence capturing semantic alignment and redundancy. Acceptability is modeled as a latent variable inferred from observable features and calibrated supervision, enabling uncertainty-aware validation decisions. The framework further incorporates dataset-level quality functionals that capture coverage, diversity, and non-redundancy, extending validation beyond individual samples to global dataset fitness. By repositioning synthetic data as calibration substrates for estimating acceptability under limited ground truth, KnowGen provides a principled foundation for constructing verifiable synthetic datasets and advances the study of data and information quality as an inferential, auditable process.
url
Article Landing PageView
Published (Version of record) Open

Metrics

1 Record Views

Details

Logo image