This research addresses the task of automatic keyword extraction, focusing on processing Russian-language texts in the field of logistics. The study identified a lack of ready-made terminological systems that accumulate the conceptual framework of the subject area – namely, the information support of warehouse logistics, distribution, and supply. To solve this problem, contemporary hybrid approaches that combine statistical and semantic methods to overcome limitations associated with the morphological complexity of the Russian language were investigated. As a result, it was determined that a combination of the BM25 statistical method and the RuBERT semantic model is the most promising for solving the stated task. A hybrid algorithm integrating the assessment of statistical relevance and the semantic proximity of terms to a given subject area has been developed. The algorithm’s mathematical model’s weighting coefficients and heuristics are justified, taking into account the specifics of the terminology. The algorithm was implemented in Python, integrating morphological analysis tools (Pymorphy2) and semantic modeling tools (RuBERT). The results of an experiment demonstrating the solution’s high efficiency are presented. The results obtained enable the automation of the formation and enrichment of the subject area’s terminological system, which opens up possibilities for its application in the ontological modeling of WMS-class software systems.