B0917
Title: Bayesian inference for partial orders
Authors: Kate Lee - The University of Auckland (New Zealand) [presenting]
Abstract: The focus is on the underlying social rank-order which constrains the ranks within lists and the order is not necessarily a complete order. The social rank order is not necessarily homogeneous in time either. The goal is to estimate the evidence for evolving social order by evolving partial order over time. Lists are uniform linear extensions of the underlying partial order in the model. We specify a marginally consistent prior stochastic process over partial orders, driven by a multivariate latent stochastic process with a hyperparameter controlling the partial order depth distribution.
