Reconstructing reliable length composition data is critical for applying length-based age-structured stock assessment models, particularly in data-limited situations. This study suggests an approach to reconstruct the length composition of small yellow croaker (Larimichthys polyactis) caught by the offshore gillnet fishery in Korean waters where existing length data often lack temporal coverage and representativeness. To address this issue, measured data by size-specific boxes were integrated with auction-based landing data. Using the number of individuals per size-specific box as a proxy, a nonparametric bootstrap resampling method was applied to generate probabilistic length composition estimates. Normality tests and ANOVA revealed non-normal and non-linear patterns in data, supporting the use of nonparametric methods and the classification of size-specific boxes. Approximately 130,000 boxes (30 million individuals), including those with partial missing data, were analyzed using an automated R-based workflow, and a hot-deck method was used for imputation. The reconstructed length compositions showed stable distributions with low standard errors, indicating improved representativeness compared to the raw data. This approach provides a practical framework for generating reliable length composition inputs to enable the application of length-based age-structured stock assessment models, particularly when age composition data are limited or poor.