We present a simple and computationally inexpensive strategy, to first estimate the average segment-orientation order parameter and then, based on the maximum-entropy method, extract distributions of this order parameter from the conventionally measured H-2 NMR spectra. The proposed methodology allows for a more complete characterization of the segmental orientation behavior as compared to solely measuring the observed quadrupolar splittings. The latter approach only quantifies the segment orientation due to local excluded volume interactions, and does not account for the orientation arising from the network topology that is manifested through the width of the spectral "wings". The application of the proposed strategy can be especially advantageous in polymer networks exhibiting highly heterogeneous segment-orientation responses (e.g., bimodal networks) and complex H-2 NMR spectrum lineshapes. The methodology is validated with both molecular simulation and experimental results as well as applied to complex spectral data of a PDMS sample.