Issue-Specific Polarization and Cohesion in a Multi-Party Legislature: Integrating the Latent Space Item Response Model with Topic-Based Regression

We develop a one-stage Bayesian framework for quantifying issue-specific legislative alignment in multi-party systems. The approach integrates a Latent Space Item Response Model (LSIRM), which embeds legislators and bills in a shared Euclidean space, with Bayesian beta regression using text-derived topic proportions as bill-level covariates. This yields posterior distributions of legislator- and issue-specific coefficients, enabling coherent comparison of polarization and cohesion across policy domains. Uncertainty is propagated through a one-stage MCMC sampler that jointly updates the latent-space and regression components. Application to the 17th Korean National Assembly reveals substantial heterogeneity in partisan conflict: fiscal domains such as Taxation and Grants and Local Government Budget show sharp polarization with tight within-party clustering, whereas Armed Services, Patriots, and Veterans exhibits weak party structuring and greater intra-party variability. The Democratic Labor Party (DLP) forms a coherent and distinct cluster on several issues even where the two major parties are not strongly polarized, confirming that important dimensions of legislative conflict are not captured by a single left-right ordering. The framework provides a principled tool for analyzing issue-structured voting behavior in legislatures where one-dimensional ideal point models yield unreliable estimates.