In this article, we construct empirical likelihood (EL)-weighted estimators of linear functionals of a probability measure in the presence of side information. Motivated by nuisance parameters in semiparametric models with possibly infinite dimensions, we consider the use of estimated constraint functions and allow the number of constraints to grow with the sample size. We study the asymptotic properties and efficiency gains. The results are used to construct improved estimators of parameters in structural equation models. The EL-weighted estimators of parameters are shown to have reduced variances in a SEM in the presence of side information of stochastic independence of the random error and random covariate. Some simulation results on efficiency gain are reported.