Various methods have been proposed to study internode interactions, but several deficiencies have been noted in certain concepts in the literature. This study introduces Algebraic Invariant Multilayer-Network Edge Regression (AIMNER), released as a Stata package to address these deficiencies. The flow and interaction templates apply the competing-destinations principle of spatial interaction theory, computing each edge as a share of the source total so that the alternatives enter every edge weight. The attribute template applies the compositional log-ratio, which compares the two endpoint attributes in relative rather than absolute terms. Since the edges are homogeneous of degree zero, they are algebraically invariant to uniform positive multiplicative macro shocks. The property is confined to this class of shocks: it does not cover heterogeneous regional shocks, asymmetric corridor shocks, nonlinear structural breaks, or channel-specific measurement error. The closed-form structure allows the framework to scale to large networks. The framework preserves the heterogeneity of each source-destination edge. The recalculation structure over periods, which includes all interactions within the closed interaction set of network edges, makes the framework applicable to dynamic networks. Since the framework operates at the feature level, it is not dependent on any estimator. The chosen estimator may use instruments for identification and clustered standard errors for inference. Together, these properties make AIMNER a feature-level multi-source data fusion framework that is, in principle, transferable to other systems with directed flows and node attributes. The framework is tested in the context of interprovincial job-seeking migration within Türkiye. Because the evidence comes from a single relational system, transfer to other networked processes is stated as a potential application rather than a demonstrated result. The results obtained in the application are consistent with the literature, showing that birthplace-based social ties are the dominant positive channel. At the same time, unemployment and educational differences function as attribute-based push-pull constraints. In the non-absorbed instrumental variables (IV) benchmark specification, the highway-distance coefficient is negative and smaller in absolute magnitude than the birthplace coefficient.
